| zdr |
Well, I could not resist building one myself, so here it is. I used Audiosector boards for amp am ps, and built bridged version with 4780. You can see that I took this project seriously, I even have a soft start circut for 2 X 200VA toroids. PS is with 4X 10000uF, MUR860 plus 1000uF Silmic near the circuit.
And ther result? Quite lousy. Compared to my Vincent SV-236 which goes for less than 1000eur, sound of this amp does not even justify the cost of the parts. I would have been (much) better off with second hand Nad 3020 from eBay.
Now, I am quite puzzled with all the good reviews on the net, so I will do some tweaking, adding some preamp and some other stuff. Any suggestions appreciated. |
|
|
| analog_sa |
| Ime GCs are extremely sensitive towards the load they drive, much more than the datasheet indicates. Some combinations just don't seem to work. What are your speakers? |
|
|
| Peter Daniel |
| quote: | Originally posted by zdr
You can see that I took this project seriously, I even have a soft start circut for 2 X 200VA toroids. PS is with 4X 10000uF,.... plus 1000uF Silmic near the circuit.
.....Any suggestions appreciated.
|
I would start with removing those additional "improvements" you so eagerly introduced, probably without listening to the amp in a basic version first.
What is your source and preamp? What speakers? |
|
|
| rabbitz |
For the LM4780, I found a preamp to be a must. You want to try the LM4780 first as described on Peter's site. I ended up with the plain stereo version which is not bridged or run in parallel and it's very sweet.
I do however prefer the LM3875 and that's the only amp I've ever built that gets away without having a pre. All my others, AKSA 55N+, GB150, NX150, P101 etc sound better with an active pre. |
|
|
| BWRX |
Hi zdr. If you post more details about your setup before shooting it down you'll likely get more constructive replies.
Have your amps had much playing time? What supply voltages are you using? |
|
|
| yusuf |
| quote: | Originally posted by rabbitz
For the LM4780, I found a preamp to be a must. |
Can you elaborate why preamp is MUST unless input signal is very low?
Thanks
Yusuf |
|
|
| PeteM |
The Vincent SV-236 appears to receive good reviews also it's an hybrid amp with a tube input stage.
http://www.lefflershifi.com/Library...cent/SV-236.htm
There will be no doubt it will sound very different to a GC for better or worse, it maybe the GC just can not compete with the current demand of your speakers.
Speaker sensitivity is not the only key to good sound, I use to use GC's on B&W CDM1 NT's, now I have higher sensitivity full range speakers. I'm sorry to say the GC just does not cut the mustard like a 10 watt Class A SS amp or my much low powered tube, yet the B&W GC combo worked a treat. |
|
|
| AndrewT |
| quote: | quote:
Originally posted by rabbitz
For the LM4780, I found a preamp to be a must.
quote:
Originally posted by yusuf
Can you elaborate why preamp is MUST unless input signal is very low? |
Hi,
an appropriate pre-amp will present a low source resistance to the power amp input.
Without the low Rs, the volume control will become a variable filter when it interacts with the filter components at the power amp input.
Additionally, a variable Rs will affect the output offset, causing variations in the DC fed to the speaker (when the DC blocking cap is omitted from the input). |
|
|
| zdr |
I am driving ProAclone 2.5, quite non-sensitive speaker. There are no big improvements, no preamp at all. Softstart circuit is on only for few seconds after the startup to prevent fuses to blow in my house due to high current peaks during startup. I ran it in for few days without a break. I yet have to listen more carefully and do A-B comparison, but the sound, even without direct comparison to Vincent, can be described as dull, closed in, with exaggerated boomy, lose mid bass and impotent low end. I am using 10K alps volume pot, with no cap at the input.
If I could attach a photo, you could see that it's a very tidy layout. It worked from the first second, and no funny noises. |
|
|
| Peter Daniel |
| quote: | Originally posted by zdr
I am driving ProAclone 2.5, quite non-sensitive speaker. There are no big improvements, no preamp at all. Softstart circuit is on only for few seconds after the startup to prevent fuses to blow in my house due to high current peaks during startup. I ran it in for few days without a break. I yet have to listen more carefully and do A-B comparison, but the sound, even without direct comparison to Vincent, can be described as dull, closed in, with exaggerated boomy, lose mid bass and impotent low end. I am using 10K alps volume pot, with no cap at the input.
If I could attach a photo, you could see that it's a very tidy layout. It worked from the first second, and no funny noises. |
First of all, ProAcs are one of few speakers that seem not to work well with GC (based on feedback I get from people using such combinations).
What you describe as "dull, closed in, with exaggerated boomy, lose mid bass and impotent low end" is exactly the effect I was getting with increased filter capacitance. Once again, those amps work best with no more than 2200uF of filtering, but to appreaciated it, you need more friendly speakers. GC is not a universal amp and it requires proper equipment matching/selection.
You mentioned using bridged version. How exactly is it achieved with your source and Alps pot? |
|
|
| analog_sa |
| quote: | Originally posted by Peter Daniel
First of all, ProAcs are one of few speakers that seem not to work well with GC (based on feedback I get from people using such combinations).
|
An interesting observation. I have a pair of seldom used ProAc Studio which sounded thin and anaemic driven from a low-cap GC. Yet, they are very acceptable with even a low-power tube amp.
What can the GC intolerance be due to? Would a parallel GC fare better? |
|
|
| lieven |
I live in Vilvoorde, got a 3875, peter's parts and 2200 panasonics, nobel pot. psu is 300 va /15v ~ in separate housing with 470 microf.
It's sounding as good as my accuphase E405 on JBL 4343B. |
|
|
| Peter Daniel |
| quote: | Originally posted by analog_sa
What can the GC intolerance be due to? Would a parallel GC fare better? |
It's really hard to say. From my experience a parallel version is hardly a solution or an improvement.
In some systems low cap GC can provide exceptional bass. For instance, in a system described here: http://www.6moons.com/audioreviews/...ctric/300b.html my low-cap monoblocks (small boxes beside equipment stand in a picture) are used to power 30-foot long straight bass horns and from what I heard, darTZeel has been also considered ;) |
|
|
| zdr |
Sorry, it's parallel, always confusing those two terms. So the one that should give more current at the output.
Pot is used with slider to lm4780 inputs. I could remove 10000uF easily and leave 1000uF only. Not sure how that can help, but I can try. |
|
|
| zdr |
| Sorry, forgot to add - supply lines are +-34V. |
|
|
| Peter Daniel |
| quote: | Originally posted by zdr
I could remove 10000uF easily and leave 1000uF only. Not sure how that can help, but I can try. |
It will be certainly interesting to hear your observations in that regard, but again, with those speakers it's hard to predict the results. One thing though, a good preamp may bring some improvements. |
|
|
| lieven |
could it be supply voltage is too high, got some 22 and 15v tranfos. Maybe current is limited because of that.
damping factor then falls below zero.
It's been a long time since I looked worked on it, but working near the 70v limit on a difficult load...
more torque, less horsepower
anyone a winisd file on proac 2.5 load seen by amp? |
|
|
| ttan98 |
If by reducing the power supply capacitance does not help in changing the sound quality you may like to try this simple change.
I assume your input impedance to your opamp is 22K, and directly couple to your 10K pot. I like you to put a capacitor(polypropylene or foil type), say 1 or 2.2uf for these 2 reasons:
1. reduce the DC interactions between the pot and the input to the opamp hence affecting the input bias current.
2. any DC output from your source may cause damage to your speaker(assume your opamp is direct coupled).
this will improve the sound. |
|
|
| rabbitz |
| quote: | Originally posted by yusuf
Can you elaborate why preamp is MUST unless input signal is very low?
Thanks
Yusuf |
AndrewT has given an excellent explanation.
An active pre sorts out the impedance matching that goes nuts with passive pre's, acts as a buffer, drives the source cable. As I said, I have only ever heard one amp that did not benefit from the use of an active pre and that one has the pot in with the power amp. |
|
|
| jduffy |
It's funny the different results people get, particularly when they seem to put more money into a GC. I built Nuuk's basic 3875 using surplus parts and am truly amazed at the sound. It's easily the equivalent of a good SS amp. I've only begun to tweak a few bits here and there. I like it as it's the perpetual "work in progress."
For preamps I use a tube linestage I built, an old Luxman preamp, or none at all. Very happy regardless of the setup. My speakers might be helping though as they're homebuilt bookshelfs that I made using Fostex drivers. Great sound for classical music. Listening to a Beethoven concerto as I type this. Just outstanding.
Good luck and I hope you sort out your GC. It should be worth the effort. |
|
|
| yusuf |
| quote: | Originally posted by Peter Daniel
What you describe as "dull, closed in, with exaggerated boomy, lose mid bass and impotent low end" is exactly the effect I was getting with increased filter capacitance. Once again, those amps work best with no more than 2200uF of filtering, but to appreaciated it, you need more friendly speakers. |
Peter, more explaination will be helpful. Since those are power line filters, I would think bigger is better. But since you have so much experience in gailcone, I am sure there must be a reason for your above statement, though I failed to undetstand why large filter capacitor would result in boomy or lose bass. In fact, I been hearing exactly opposite in few threads here.
| quote: | Originally posted by Peter Daniel
GC is not a universal amp and it requires proper equipment matching/selection. |
Can you give some guildline on matching criterion.
Thanks
Yusuf |
|
|
| zdr |
| quote: | Originally posted by lieven
could it be supply voltage is too high, got some 22 and 15v tranfos. Maybe current is limited because of that.
damping factor then falls below zero.
It's been a long time since I looked worked on it, but working near the 70v limit on a difficult load...
more torque, less horsepower
anyone a winisd file on proac 2.5 load seen by amp? |
That is true, but don't forget I have two 200VA toroids, so I think I can afford 34V rails... |
|
|
| rabbitz |
As you increase your rail voltage, the low impedance driving capabilities decrease. I'm not familiar with the Proac but I'm sure the impedance will be reasonably low.
To cover low impedance speakers on my LM4780, I use 26VDC rails and use 100uF caps at the chip and snubbered 10,000uF on the power supply.
Have a good read of the datasheet and info on Carlos's snubbered supply. Peter's power supply PCBs has allowances for the snubbered supply.
National Semi have a good design guide (used to be a spreadsheet). Have a look at what rail voltages are acceptable for 4R loads in the graph and anything over 30VDC is not good. |
|
|
| rabbitz |
| That graph only gives part of the picture as you need to see the others in the datasheet for 6R and 8R. You have to make the rail voltage suit the speaker. Here's the one for the LM3875 which shows as rail voltage goes up, low impedance capabilities fall. |
|
|
| ttan98 |
"As you increase your rail voltage, the low impedance driving capabilities decrease"
If I remember correctly, Output impedance is not a function of power supply, it depends on loop gain of the opamp which is plenty for this design. As the power supply increases, the power output increase so is distortion and it does NOT mean the output impedance increases.
For what I read here LM3875 does not have enough current to drive low load eg 4 ohms and lower. LM4870 is a dual LM3886 and each LM3886 can drive a 4 ohm load. Daniel and others believe so, in other postings I read. |
|
|
| baboon81 |
| I have to agree with rabbitz and AndrewT. I build two mono blocks using the LM3876 and with out any preamp it just didnt "sound right". Then I built a simple buffer (from Rod Elliots site) using the OPA2134 op amps with a gain of 2. After connecting the 100k pot to the buffer then the buffer to the lm3876 amp, the amp sounded much "better". |
|
|
| owen |
Sadly, you are hearing the limitations of your 'cloned' speakers.
I've been looking at the response curve of the tweeter, and in a number of 2.5 clones that are on the web (in the main, using the same tweeter) if its is like the one designed by Troels, then there is a very visible drop in output above 15KHz. That would sound 'closed' and 'shut in', especially if no buffer is used with long interconnects - you are then placing another low pass filter in the audio stream!
http://www.troelsgravesen.dk/download/2.5_Clone.pdf
The loose and boomy bass is because the cabinet size is acoustically tuned (i.e mainly by ear, but also has reasonable measurements), and may well be exciting room modes, especially if they are close to a rear wall (excites the primary modes more).
In a transformer coupled amp, the power output availble for the bass is reduced with reduction in frequency because of the transformer.
This is not the case with the gainclone. It's flat. If it needs to deliver 50 Watts at 10Hz, it'll do it for as long as needed assuming a correctly sized heat sink. These amps are used to derive a 50 Watt power supply for turntables, so they can drive the LF very well. The increased cap powersupply means that the amp can deliver bass transients in excess of that stated output!
So. in summary, you will hear a different presentation with a lower cap supply (that should help with the bass situation), and you will need a pre-amp to ensure that the input of the gainclone is properly driven with this speaker. These two changes will make a substantial difference to the sound.
Reducing the capacity in the powersupply will reduce the inrush current, and that means the 'soft start' circuit can be binned too!
So, in order, reduce the psu capacity (unsolder the 10k uF caps you have there), take out the 'soft start', and use a preamp/buffer before your gainclone. Then see what it can do.
Owen |
|
|
| juergenk |
| quote: | | Without the low Rs, the volume control will become a variable filter when it interacts with the filter components at the power amp input. |
if even the audiosector amps have not any such components at the input? |
|
|
| owen |
YES!
There will be a capacitance at the input that will be a direct product of the circuit used in the chip.
There will also be capacitance in the cables too, so even if there isnt a direct input capacitor, there will be a 'virtual' capacitor in series.
Because it's not seen, it is often forgotten...
Owen |
|
|
| AndrewT |
| quote: | Originally posted by ttan98
"As you increase your rail voltage, the low impedance driving capabilities decrease"
If I remember correctly, Output impedance is not a function of power supply, it depends on loop gain of the opamp which is plenty for this design. As the power supply increases, the power output increase so is distortion and it does NOT mean the output impedance increases.
For what I read here LM3875 does not have enough current to drive low load eg 4 ohms and lower. LM4870 is a dual LM3886 and each LM3886 can drive a 4 ohm load. Daniel and others believe so, in other postings I read. |
output impedance and low impedance driving capability are two completely different concepts.
The chipamps have severe problems sourcing adequate current into low impedance loads.
It shows in the extracted graphs posted by Ttan98.
if one converts the output power capability of the chosen chipamp into the output Vpk and Ipk, then one begins to see the difficulty.
What is often overlooked is the effect a reactive load has on the current demand from and dissipation of the chipamp.
4ohm speakers are a demanding load and a severe 4ohm load with a load phase angle approaching 60degrees will trouble just about any single chipamp. Paralleling helps solve the problem. Lower supply rail voltage helps as well. Neither of these solutions lowers the output impedance. The first increases the available current, the second lowers the Ipk supplied to the load. |
|
|
| AndrewT |
| quote: | Originally posted by owen
Sadly, you are hearing the limitations of your 'cloned' speakers.
I've been looking at the response curve of the tweeter, and in a number of 2.5 clones that are on the web (in the main, using the same tweeter) if its is like the one designed by Troels, then there is a very visible drop in output above 15KHz. That would sound 'closed' and 'shut in', especially if no buffer is used with long interconnects - you are then placing another low pass filter in the audio stream!
http://www.troelsgravesen.dk/download/2.5_Clone.pdf
The loose and boomy bass is because the cabinet size is acoustically tuned (i.e mainly by ear, but also has reasonable measurements), and may well be exciting room modes, especially if they are close to a rear wall (excites the primary modes more).
In a transformer coupled amp, the power output availble for the bass is reduced with reduction in frequency because of the transformer.
This is not the case with the gainclone. It's flat. If it needs to deliver 50 Watts at 10Hz, it'll do it for as long as needed assuming a correctly sized heat sink. These amps are used to derive a 50 Watt power supply for turntables, so they can drive the LF very well. The increased cap powersupply means that the amp can deliver bass transients in excess of that stated output!
So. in summary, you will hear a different presentation with a lower cap supply (that should help with the bass situation), and you will need a pre-amp to ensure that the input of the gainclone is properly driven with this speaker. These two changes will make a substantial difference to the sound.
Reducing the capacity in the powersupply will reduce the inrush current, and that means the 'soft start' circuit can be binned too!
So, in order, reduce the psu capacity (unsolder the 10k uF caps you have there), take out the 'soft start', and use a preamp/buffer before your gainclone. Then see what it can do.
Owen |
| quote: | Originally posted by owen
YES!
There will be a capacitance at the input that will be a direct product of the circuit used in the chip.
There will also be capacitance in the cables too, so even if there isnt a direct input capacitor, there will be a 'virtual' capacitor in series.
Because it's not seen, it is often forgotten...
Owen |
Hi Owen,
if I understand your postings correctly then I fear you have mis-understood the science/electronics/maths.
The result is that some of your conclusions are flawed.
Do you want to Email me? |
|
|
| juergenk |
ok!
But these parasitic components are part of any amp, not only the natsemi series.
Why are the gainclones considered something special?
Regards |
|
|
| analog_sa |
| quote: | Originally posted by owen
In a transformer coupled amp, the power output availble for the bass is reduced with reduction in frequency because of the transformer.
|
Which amp is transformer coupled? The Vincent? Really?
Even if it was at what frequency would there be anomalies? 5Hz?
Most reviewers mention the bass response as the amps best feature. |
|
|
| owen |
Andrew,
Thanks for the offer..
I suggest that a close look at the 'budget' seas tweeter commonly used for these 'clones' - they can have up to 6db drop at 20KHz dependant on the crossover topology used.
I have looked at the 'vincent' amps, and from what I can see they are a hybrid bipolar output stage with valve buffer, biased to give a higher wattage class A output than a standard A-B stage.
My comments specifically for the proaclones (and the proacs) relates to the use of an amplifier that has a transformer output (typical valve amp, but has been used with SS too).
These have a reduced wattage output at LF due to transformer core saturation, that can mask the true LF acoustic output of the speaker, and in turn help minimise the excitation of room modes. That also ignores the difference in damping factor of the amplifier...
The PDF links to Troels designing the clone itself, and to paraphrase - the cabinet volume is approximately 67 litres, but the simulations suggest 42 litres on page 10.
The fact that the speaker output is circa 10 ohms for Troels design, mean the rail voltages should be ok, and the phase shifts are + - 35 degrees and have smooth changes above 100 Hz, as would be expected of a stuffed transmission line design.
Juergen,
The input capacitance (as opposed to the input capacitor) requires current to drive it, and it is an integral part of any amplifier stage. It has a value that is a part of describing the input impedance.
Impedance is different to resistance, in that is varies with frequency, and is electrically described using an equivalent circuit comprised of an inductor, resistor, and capacitor.
As the impedance varies, the output of the buffer must be able to vary the current to maintain the voltage (as these are voltage amplifiers) at the main amplification stage, otherwise the voltage will drop. This voltage 'drop' at that frequency, will be audible, because in effect the 'volume' has been lowered.
Now onto the gainclone. In the gainclone, it can be configured to have a relatively high input impedance, so these effects are reduced. This lead many to use (and lots of people like them like this) just a simple potentiometer as the attenuator in the circuit.
However, this means that you are dependant on the quality of the source buffer to be able to maintain the current to drive not only the capacitance of the cabling, but into the input impedance of the amplifier. It must also be able to do this through the resistance of the attenuator too.
The capacitance of the cable, and the input impedance are in parallel, so will be summed! This is what produces the 6db/octave low pass first order filter to your amp, and is what produces the characteristic HF roll-off.
This 'problem' is obviously exacerbated by long interconnects (source - attenuator - power amp is typically 2 meters!), and a weedy output buffer on your source.
The most dramatic demonstration of this is in a badly designed triode amplifier, where the miller capacitance is large, and the input stage is current poor - for example a 12ax7 driving a 6c33.
To equate that to your system, the 12ax7 is equivalent to the output from your source, and the 6c33 is equivalent to your power amp. In a valve amp, in those cases, you insert a 'driver' stage to provide the necessary current - and with a gainclone, an active preamp, or buffer does exactly the same job.
It ensures that there is sufficient current to maintain linearity across the audible spectrum.
Owen |
|
|
| owen |
Proacs are noted as valve(tube) amp friendly, because of benign impedance curves, higher impedance overall than many designs. They are also engineered to have a tightly managed frequency response, and offer good LF extension. This LF extension is rarely fully encountered with a valve amp, simply because the output transformer core cannot swing enough magnetic flux (it in effect goes into magnetic 'clipping'), and so output falls below a common 40 or so Hz, but there are exceptions to this (OTLs, and massively engineered designs).
So reduced LF output from the amp, means reduce LF output from the speakers, that means less energy available to excite LF room modes.
Ok?
Owen |
|
|
| ttan98 |
Please read my posting I DID NOT SAY the follow:
"As you increase your rail voltage, the low impedance driving capabilities decrease"
rabbitz did.
Also, you said
"It shows in the extracted graphs posted by Ttan98."
Rabbitz posted it I did not.
So please read my posting carefully.
Each amp has 0.1ohm connected to it if both of them are connected in parallel, that increase the output impedance by a small amount and decrease the driving ability. |
|
|
| AndrewT |
Hi Rabbitz, Ttan98 and all,
sorry for the erroneous attribution. I substituted Ttan98 for Rabbitz. Got myself a bit mixed up.
The general meaning is hopefully still understood.
Ttan98 has got it wrong.
I did read Ttan98's posting and other than the wrong attribution my message is the same. |
|
|
| owen |
Andrew,
No worries...
Owen |
|
|
| rabbitz |
Yes Andrew, I knew what you meant.
Ttan98, in my illustration I wasn't referring to the output impedance of the amp but to it's ability to drive low impedance loads, hence the graphs from the datasheet. Although the LM3875 is not relevant to this, it does show how these chips react to voltage vs impedance load. It's also on the LM4780 datasheet but on separate graphs.
I've played around with Peter's LM4780 kit in stereo and parallel mode and ended up with a stereo module which is used in a workshop test system. Tests are done often with low impedance speakers and I can tell you with higher voltage rails the chip is not the happiest camper under these loads. That's why I dropped the rail voltage so it can accommodate all the speakers I try and a side benefit is running much cooler which is important as the workshop is not air conditioned on hot days. I lose power by doing this but it's working well within it's limitations. |
|
|
| rabbitz |
Here's some examples with the Overture Design Guide 15.xls.
This one's 34V into 4R and it's putting some heat out. |
|
|
| rabbitz |
| Now 34V with 8R.... all is well. |
|
|
| rabbitz |
| ... and what I ended up doing was 25V into 4R. This keeps the temperature down and still allows 4R capabilities. This is for a stereo module only BTW. |
|
|
| ! |
| quote: | Originally posted by rabbitz
... and what I ended up doing was 25V into 4R. This keeps the temperature down and still allows 4R capabilities. This is for a stereo module only BTW. |
You seem to imply heat is a problem but rather it's inherant in having a higher power capability. It seems you merely need a bigger heatsink if you want that output. While I have no argument that not everyone needs to try to squeeze 50W out of the chip, citing higher voltage as bad, when there is no thermal cutout observed, is a bit premature. |
|
|
| AndrewT |
Hi,
using a supply voltage selected to suit the load impedance is crucial.
The datasheet design process confirms that lower impedances need lower voltages.
The datasheet also confirms that as power increases so does the need for a bigger sink. But there is a limit, the Rth c-s of the single chip. It becomes the bottleneck. This effect is worse with lower impedances.
One cannot maintain supply voltage as the load impedance drops by using ever larger heatsinks. |
|
|
| rabbitz |
| quote: | Originally posted by !
You seem to imply heat is a problem but rather it's inherant in having a higher power capability. It seems you merely need a bigger heatsink if you want that output. While I have no argument that not everyone needs to try to squeeze 50W out of the chip, citing higher voltage as bad, when there is no thermal cutout observed, is a bit premature. |
I'm not saying higher voltages are bad as they work very well on the right load. I've just found it easier to tailor the rail voltage to suit the speaker and in my application could have some nasty impedance loads.
AndrewT has nailed it and I was passing on my experiences with the LM4780 configured in stereo. You can only throw so much heatsinking at the chip.
I have no idea how much of this applies to the chip in parallel and bridged mode as I've only used it in stereo.
This is the recommended heatsink size for 34V + 4R according to the National Semi design guide and it's saying it's too big for the chip. |
|
|
| Nordic |
| On another level, the actual chip packageing has its own thermal bottleneck... the silicon chip inside can heat up at a faster rate than the case/sink junction can get the sink to respond... so at some point, everything will burn off even before the heat reaches the sink. At that point it does not matter what size heatsink you use...it will burn. |
|
|
| AndrewT |
| quote: | Originally posted by Nordic
On another level, the actual chip packageing has its own thermal bottleneck... the silicon chip inside can heat up at a faster rate than the case/sink junction can get the sink to respond... so at some point, everything will burn off even before the heat reaches the sink. At that point it does not matter what size heatsink you use...it will burn. |
no.
Thermal inertia applies certainly, but so too does transient survivability. |
|
|
| ! |
| quote: | Originally posted by AndrewT
Hi,
using a supply voltage selected to suit the load impedance is crucial.
The datasheet design process confirms that lower impedances need lower voltages.
The datasheet also confirms that as power increases so does the need for a bigger sink. But there is a limit, the Rth c-s of the single chip. It becomes the bottleneck. This effect is worse with lower impedances.
One cannot maintain supply voltage as the load impedance drops by using ever larger heatsinks. |
Certainly voltage increase can't be infinite. That is a higher threshold, different than looking at a graph that shows what happens when a heatsink interface or 'sink itself is insufficient to keep it cool enough. Remember even then, a datasheet is a guaranteed minimum, worst-case scenario.
I think it would be fairer to say that given a good enough heatsink, voltage above a certain point doesn't have a purpose, but there is still a range that could be used without expectation of problems, that use of an existing supply-components is weighed against the actual voltage and performance of the amp as related to the heat generated. Plus, we're playing music, not a sine wave, so it is possible to have momentary peaks that don't trip the Spike protection because the average heat is lower, momentary heatsinking effectiveness has a lot to do with the 'sink and die temp up to that moment.
We can also see this from some passages in the app notes like
| quote: | there are two output power specification guarantees: 60W (min) into a 4Ω load using ±28V supplies and 50W(typ) into an 8Ω load from ±35V supplies. Using these two conditions and the theoretical
maximum power dissipation equation ...
... These results show that the IC can handle a maximum of
≈ 40W of continuous power dissipation without SPiKe Protection
being turned on under continuous sinusoidal input
with proper heat sinking.
... Your electrical design parameters and thermal management
may be different, changing the achievable results. As always,
lab testing is recommended to verify any solution.
[AN-1192, pg. 3,4] |
I for one would as soon have an amp capable of higher momentary peaks to keep dynamics, even if it couldn't play a sine wave at that level indefinitely, but I don't mean to suggest completely abandoning voltage considerations, just that there is a little latitude. |
|
|
| Nordic |
| There certainly are members who exceed voltage limits by a few volts, most of them very pleased.... high impedence loads off course. |
|
|
| Leolabs |
| Guys,did zdr mention his source and output level,yet????? |
|
|
| Graydon |
Hi Guys,
I built a pair of the 3886 chip amp modules from Chipamp.com, each with its own power supply. (the dual mono setup) I usd no volume pots or anything like that. I use a Harmon Kardon receiver for a preamp which feeds the Chipamp directly.
Initially, I had a bunch of 12VAC transformers on the shelf so I used two of those for the power supply. You can see these in the photos on my blog. That gave me a fairly low B+ but it worked well. I did all of my initial testing on a pair of really nice speakers that were built by a DIYer named Keith Kidder. More details at my blog.
A couple weeks after I finished my Chipamp, I finally received a pair of toroid trannys with dual 22V secondaries which I swapped in. My contribution to this thread is that my first observation besides the reduction of mechanical noise with the toroids was that the bass response with the higher voltage toroids seemed to be more sterile. Sure, the amp is obviously more powerful and the bass is adequate, but with the 12V supply trannys, the bass was really rich and seemed to resonate in the room at down near 25Hz. The comments that you guys are making regarding voltage and response seem to imply that maybe my observation is warranted. I'm still digesting all that is being posted in this thread to try to figure it out.
http://gdsamps.wordpress.com/chip-a...chassis-60-wpc/ |
|
|
| analog_sa |
| quote: | Originally posted by Graydon
a pair of toroid trannys |
Any conclusions drawn out of multi-variable experiments are pretty meaningless. Yes, it's possible that the chips play better at lower voltage. It is also possible that you prefer the sound of EI transformers.
Ime PS transformers sound surprisingly different and i really wish i knew what makes some sound better. |
|
|
| zdr |
| quote: | Originally posted by Leolabs
Guys,did zdr mention his source and output level,yet????? |
Nope, but here it is :)
The sources are the same for both setups - Original CD2008 and PC lossless->DAC Diyeden SVDAC01 Plus with modified output (OPA627 and no nasty caps in series).
I still did not get the chance to test it without 4X10000uF in PS. I did a test with caps on the input (Audyn CAP 4,7uF, 400V) and did not make much difference (if any at all). DC was zero across the caps so out they went again.
I did make another set of speakers though - HATT MKII, completely different animal, and will be doing some A-B comparisons soon. Vincent also developed a problem in preamp's right channel so I had to fix that too (bad solder joint under a tube).
All those are the reasons I was quiet for a while. I will be doing some tests soon and will play it again to my wife - she can hear much more differences. |
|
|
| AndrewT |
| The lengths some will go to to please a wife. Any wife? |
|
|
| Alien8 |
zdr
I have some proac clones too, which I'm running with mauro's rev C (twisted pear audio kits). Running it of a preamp (pga2310) they do sound fine to me ! Though that's something completely different from a gainclone.
You're welcome to come by to compare, only 30 minutes away from you in Leuven (except for traffic)
A8 |
|
|
| zdr |
| Thanks for the offer, I might accept it since nothing that was suggested made a difference. Don't get me wrong, it sounds ok until I compare it to Vincent SV-236. I tried now without 10000uF caps and with quality caps on the input - no difference. I even tried with completely different speakers (HATT MkIII) and still it's a mediocre, ~300 eur sound. My expectations were somewhere in 2000K area. I maybe expected too much? |
|
|
| Alien8 |
You're wellcome, pm me to see when it fits.
Do you have a way to measure the gainclones freq response ?
I have heard gainclones on electrostats, no problem though they were certainly running hot ! Apogees are not very sensitive speakers and definitely not an easy load so...
Another option is to turn it in bridge mode to make a subwoofer amp ;-) That'll surely beat a 300 euro sub amp !
A8 |
|
|
| rabbitz |
| quote: | Originally posted by zdr
My expectations were somewhere in 2000K area. I maybe expected too much? |
The LM4780 is good but not at that sort of level IMO. You have to get into AKSA territory to achieve that. I have an AKSA 55N+ as a reference and no other amp has come close. |
|
|
| yusuf |
| quote: | Originally posted by zdr
Thanks for the offer, I might accept it since nothing that was suggested made a difference. Don't get me wrong, it sounds ok until I compare it to Vincent SV-236. I tried now without 10000uF caps and with quality caps on the input - no difference. I even tried with completely different speakers (HATT MkIII) and still it's a mediocre, ~300 eur sound. My expectations were somewhere in 2000K area. I maybe expected too much? |
Additional PS filter capacitor improves the supply voltage. Hence, IMO, it should not be expected that removing 10000uf cap will imprve anything. However, downside is possible.
What really disappoints you? Isn't it faithful reproduction of input signal? Hearing test is really very subjective. |
|
|
| ! |
The problem with this user-preception-as-evaluation is we don't have a baseline.
You compare it to another amp but personally, I've never heard that amp and don't know it's particular benefits. Further what is missing is a thorough description of exactly what sound you are looking for.
Some consider accurate response, a faithful gain on the input signal to be "boring" or bland, whatever the term du jour. "Maybe" you do too, but if what you are seeking is not a mere gain on a flat response, you will have to specifically tailor the amp to your goal. Maybe a gainclone can accomplish that or maybe it can't, it depends on what the goal IS.
I will say one thing though, unless your goal is distortion (which is not necessarily a bad thing, it depends on who you ask and context), there is no justifiable reason not to use more than 1000-2200 uF capacitance. Beware of advice by those only considering a handfull of chipamp kits because they are not considering the grand scheme of chipamp power circuitry, they are only considering some crude or simple implementation as if you had no alternative at all but to have either a few giant caps or only a local resevoir that is very large.
Ignoring the possibility that you may want to color the sound for a moment, it is wrong to build a chipamp with only one 1000 cap or distant much larger caps on the PSU rails. People perpetuate that myth by misunderstanding the datasheet, that when a datasheet shows 1000uF it does NOT AT ALL MEAN the only thing in the circuit past a bridge rectifier is this lone cap per rail. What it means is, you are supposed to assume ideal and perfect power in your device right up until this point in the circuit, including a hypothetical 0.0000% ripple, and that the parts spec'd are only to deal with the ripple induced by the active device whose datasheet you are looking at.
If you don't yet have 0.000% ripple up to that point, you are not gaining anything by lowering capacitance. It is just the wrong answer unless you are trying to color sound - but maybe you are!
Anyway, once you get that ideal 0% ripple up to that point in the circuit, at that point you don't need thousands of uF of capacitance close to the chip. What you do need is at least 100uF, possibily better to move closer to 220-470uF as close to the chip as possible (except between it and chip, a reasonably quality fractional %/uF film cap for decoupling, like 0.1uF or even a pair of 0.01-0-.1uF & up to 1uF. Some will claim paralleling can cause resonance. Yes, in rare cases it can when electrically paralleled but distanced on the boards and wires, but generally the benefit outweighs the slight risk, and if the resonance happens on an circuit you design yourself and you need to eliminate it, the better resolution is not removing the cap, rather reworking the trace resistances to a minimal level.
In short, the ideal of a chipamp where you just pick one value of single cap per rail for PSU filtering/decoupling is a crazy myth. It is perpetuated by misinterpretation of the datasheet and perhaps some selling chipamp kits or finished projects that were based on the myth, but one thing it is most certainly not is some crazy alternative like "you can only have one cap, picking between some giant multi-thousand uF or one smaller electrolytic like 1000uF".
Remember the basic rules of electronics. If a chipamp designed with as near a perfect power supply system is not what you want, it's probable you don't want a chipamp at all. This also assumes you had scrutinized all other aspects of the circuit, if you are only trying to compare to one particular amp then I highly suggest you carefully compare all aspects of deviation between these two including supply voltage and current per watt output, coupling cap type and value vs impedance, and anything else that seems applicable towards the perceived differences between the two.
This doesn't change the fact that if you could only compare two basic cap reservoir and decoupling designs, one may sound different than the other. In the end, ideally you don't choose one or the other at all, you incorporate the best of both, BUT, maybe you are trying to match the amp output to the quirks of certain speakers. That is a losing battle IMO, it is better to have a faithful reproduction from the power amp then if you want color, do it with the speakers or a pre if you must.
In the end a chipamp can be made into a very reasonable solid state amp, but it leaves less room to modify to tailor the sound, So it depends on whether your goal is being able to modify, or striving for what a decent short loop solid state amp can provide. |
|
|
| janneman |
For once a well reasoned, consistent and knowledgeable post. Refreshing!
Jan Didden |
|
|
| jackinnj |
| Besides looking at the GC's with a network analyzer and THD analyzer I have actually listened to them -- usually with an Arcam DVD player or Linn LP-12 and my Adcom modified GFP565 preamp -- they do pretty nicely with everything from my Bevelled Thors, an old but mint condition AR3a's, Linn Kelidh's, Kef 104ab's, and my KEF B139 reflex spkrs -- but I can offer this observation -- if you aren't careful about the wiring you can getty some nasties on the output -- it's an opamp on stereoids, high gain around a lot of elevated voltage -- further without the correct heatsinking you can get the amplifier to "hesitate" before it goes into thermal protection -- overall they are pretty trouble-free and easy to implement and national gives you a lot of tools to work with. |
|
|
| pedroskova |
| I'm not familiar with Peter's Audosector boards, but I seem to remember his designs as being pretty minimalist in nature. It could be that, if this board doesn't contain a zobel on the output, the dissappointing outcome is the result of an oscillating chip that doesn't like the speaker cable / speaker combo being presented to it. |
|
|
| rabbitz |
There is provision for a zobel on the output on Peter's boards and if I remember parts are also supply on the LM4780.
I think that Peter does not use a zobel on his own amps even though the board allows for it. |
|
|
| zdr |
It's all about reference or baseline, if you wish - I completely agree. That's why I tried to stamp a price tag to the sound I am hearing. It's as quantitative as I can get when it comes to a sound. Vincent SV-236 is very good 25Kg class A amp, which some time ago humiliated Musical Fidelity X-150 in bass performance and transparency, right in front of my ears. It's still a Chinese that goes for well under 1000k euro.
The most noticeable deficiency of the sound from my GC is that it's pretty much out of control in higher upper-bass region - loose, elevated and clouded. Very tiresome and bothering. I did not have chance to audition it more carefully with my kid jumping around, but that one thing really stands out even with background noise and distraction. I will try to be more precise when I get next chance.
It does lousy job on two very different speakers, but consistently, so I think we could pretty much rule the load out.
I might add I am not a person with golden ears, and I really cannot hear difference between interconnects. Hell, I even struggle to hear difference between speaker cables, not to mention AC ones. I might also add that I have master's degree in electronics, so I think I did a fair job in putting it together. |
|
|
| ! |
Forgive my laziness if I overlooked your design but is it regulated? If not, well...
What output level are you auditioning when noting the lack of upper bass control?
Personally, I would use beefier wiring from the large reservoir caps and smaller caps on the chip boards, and the transformers are a little close for not even having belly bands on them, especially with the inputs right behind (but one thing at a time, isolating variables is quite important). Don't get me wrong it looks nice, but if we're hunting for a skunk we might have to poke around a bit and see what smells funny. However, I think it could be wrong trying to duplicate one amp with a different one. I say forget about the other amp you like more and just focus on what you can do to this one from this point forward. I mean, c'mon, they are not the same beast, add a tube input if you're trying to build one into being the other, if the tube had no effect it wouldn't be there.
On the other hand, I didn't mean to slight what Peter Daniel had to say about the changes you'd made from his recommended version. He has done a lot of these and that has to count for something. It could be that even if this amp does not sound as you prefer with the other amp, the result could be unique and fill a niche. Who really needs more than one amp with any given sound signature? Plus, I do still wonder if your speakers are partly to blame, that they don't exhibit this with your favorite amp could as much be a sign it lacks the flat upper-bass you now find excessive.
I also wonder if you're asking too much, comparing to a Class A amp. I don't believe at all in describing based on a price range though, when it comes to power amps that can be an arbitrary thing as much to do with marketing and volume vs market saturation as anything else, and let's face it, the engineering and construction isn't the same, the more advanced the design the smaller the market and the price skyrockets. By the same token no matter how good you managed to get a Gainclone to sound, someone else could come along and say "I describe it by price therefore gainclone must sound poor". |
|
|
| zdr |
I am back:) I built a pair of Zaph's ZD5s so I dug out my gainclone for testing purposes. I was surprised to notice that it plays much better now?? I cannot make a distinction between vincent and gainclone any more. Nothing changed unless some rats made some silent modifications, I even soldered back big PS elkos. Unfortunately I don't remember what was previous setup for A/B testing, but it's quite simple anyway. To speed up A/B switching, I am using rec out on vincent to line in on gainclone. This is not preamp out though, it's right behind the input switch.
Are my ears playing tricks on me? Have to play it to my wife, she could always here more details. |
|
|
| jackinnj |
| quote: | Originally posted by !
What output level are you auditioning when noting the lack of upper bass control?
|
When comparing amps I have observed that even a a small variation in output makes a big difference in listening impression. |
|
|
| jrockhead |
[QUOTE]Originally posted by zdr
[B]Just to have an idea about the amount of work put into this project, here's the image.
Greetings, Beautiful job of building your amp but three things jump out at me. First, wood is nice to build an amp with but it has no EMI/RFI shielding.
Second, it looks like the bottom of the heatsinks are sitting on top of the bottom wood plate. Doesn't help with cooling airflow and these amps do get warm.
Thirdly and probably the most vehemently discussed topic is did you let the amp play for at least 20 hours before any critical listening? Were any boutique caps used such as Blackgates? I love em' but they need a large break-in time.
Just let the amp play for a couple of days and see what happens. From your post you might have heard the amp settling down already though
Regards, Marcus. |
|
|
| maxlorenz |
Hi ZDR,
I was about to post about my first chipamp excellent experience when I saw your thread tittle!
I am glad that your experience is better now. Might have been hard to drive speakers + burn-in time...
Anyway, here comes my humble adventure:
I built my first ever LM3886 based chipamp with regulated supplies (+/- 29VDC). The regulators are the "PowerRegs" (BC550/560 instead of JFet) developed by Teddy Pardo @ pinkfishmedia .
http://www.pinkfishmedia.net/forum/...ead.php?t=39306
He describes the sound of his own regulated gainclone as the best amp he has heard (he even uses low value Tants near the chips; see photos), so I could not help building it.
I built it for my eldest daughter, for her to enjoy music "comme il faut" but I am considering seriously to keep it, so good it sounds...after 3 hours of warm-up it changed from "not bad" to "unbelievable". The bass is deep and powerful (my speakers are easy).
The best part is price as the stereo board with two LM3886 costs US$26 at http://eshop.diyclub.biz/index.php?c...205_150_22_203
A photo of the test setup:
http://picasaweb.google.com/maxlore...652793070188882
Teddy has an excellent regulator for low power applications, the "TeddyReg". I am in the process of changing all my regs for them.
A humble contribution from my part (at least I expect it has not been discussed here), which I have not tested yet on my amp but on opamps for DAC output, is the one consisting in opamp compensation for capacitive loading. The detailed explanation obviously escapes my limited "capacity" but useful links can be found here:
http://64.233.169.104/search?q=cache...ient=firefox-a
http://www.analog.com/Analog_Root/st...stability.html
This tweak was recommended to me by -ecdesigns-, and it is easy to implement: just solder a resistor between signal inputs (+) and (-) , whose value must be five times the value of the Feedback R, for best sound. In power amps an RC must be soldered, C being at least (?) 680nF. In one reference I found "as large as feasible" as a value. I don't know which reference is better. Resistor shall be a good metal-film R and that C shall be a film cap.
So, I don't want to hijack the thread...only wanted to share my experiences and expect some comments from you, the more experienced on gainclones...:angel:
Regards,
M |
|
|
| danielwritesbac |
| quote: | Originally posted by jrockhead
. . . Thirdly and probably the most vehemently discussed topic is did you let the amp play for at least 20 hours before any critical listening? Were any boutique caps used such as Blackgates? I love em' but they need a large break-in time.
Regards, Marcus. |
For those (like me) looking for a solution to the difference between artificial-flat response and modeled human-ear-flat response on LM chips, I'm researching the cap topic.
So far, it seems that the LM amplifiers can be brought to within tolerances for human ear documentations by using the C and Ci caps either a large signal stable ecap from Panasonic (successful but obsolete) or a purpose made audio ecap that's got quite high resolution bass, and then bypass (parallel) it with 0.01uf to 0.005uf (10nf to 5nf) film cap to get. . . the "warmpth and air" and the "wide presentation" effects. . . for you to have this option.
Other applications include support for inefficient speakers, because the new margin on response can bring useful gain settings up to 45 without overly increasing the amplifier's sonic signature (LM3875, LM1875). In testing this has worked.
Other advantages include twice the power and a change to "soft clipping" instead of the usual sudden onslaught. However, this week, I'll attempt to see if I can be specific as to the current production model #'s of widely available caps suited to the task, and let the testing fun begin. ;)
Wish me luck, and thanks for all your help. |
|
|
| garage1217 |
| quote: | Originally posted by zdr
I am back:) I built a pair of Zaph's ZD5s so I dug out my gainclone for testing purposes. I was surprised to notice that it plays much better now?? I cannot make a distinction between vincent and gainclone any more. Nothing changed unless some rats made some silent modifications, I even soldered back big PS elkos. Unfortunately I don't remember what was previous setup for A/B testing, but it's quite simple anyway. To speed up A/B switching, I am using rec out on vincent to line in on gainclone. This is not preamp out though, it's right behind the input switch.
Are my ears playing tricks on me? Have to play it to my wife, she could always here more details. |
The zd5's work very very well with the power you are feeding them. I have my 4780 dual mono "parallel" setup in the works at the very moment to drive a pair of zd5's as well :) I originally chose peters kit specifically to drive these speakers. Very glad it has worked out for you. Listen, enjoy and smile at what you have created! The ZD5's are truley magic and with a good amp behind them, magic happens. |
|
|
| BWRX |
Wow, where to start...
| quote: | Originally posted by danielwritesbac
For those (like me) looking for a solution to the difference between artificial-flat response and modeled human-ear-flat response on LM chips, I'm researching the cap topic. |
The ideal amplifier has a completely flat frequency response. Your use of "artificial-flat" is completely misleading.
I have no idea what you think "modeled human-ear-flat" response means, but it's not something that you want an amplifier to have. If you want to see what effect changing capacitors has on the frequency response of your amplifier, connect a frequency generator to your amp and measure it!
| quote: | Originally posted by danielwritesbac
So far, it seems that the LM amplifiers can be brought to within tolerances for human ear documentations... |
"human ear documentations"? Please stop spouting gibberish.
| quote: | Originally posted by danielwritesbac
Other applications include support for inefficient speakers, because the new margin on response can bring useful gain settings up to 45 without overly increasing the amplifier's sonic signature (LM3875, LM1875). In testing this has worked.
Other advantages include twice the power and a change to "soft clipping" instead of the usual sudden onslaught. However, this week, I'll attempt to see if I can be specific as to the current production model #'s of widely available caps suited to the task, and let the testing fun begin. ;) |
Increasing the gain DOES NOT increase the power of the amplifier!
There is no soft clipping behavior of the LM amps. They exhibit hard clipping like almost all solid state amps. The Spike protection is not a form of soft clipping. These are two completely different things. Please, please (!) read up and ask questions about this stuff before trying to relate your ideas to others. While your intentions are good, your recommendations and explanations are very misleading and confusing. |
|
|
| ShinOBIWAN |
| quote: | Originally posted by BWRX
I have no idea what you think "modeled human-ear-flat" response means, but it's not something that you want an amplifier to have. If you want to see what effect changing capacitors has on the frequency response of your amplifier, connect a frequency generator to your amp and measure it!
"human ear documentations"? Please stop spouting gibberish. |
I think he's talking about those fletcher/Munson equal loudness curves that show how the ear/brain interprets a known and measured flat response as being higher or lower in level depending on frequency. In a nutshell; we measure it flat but the ear/brain interprets this as non flat due to its own non linearity. To present the ear with a flat response the input signal has to be shaped. Its all psychoacoustics with not enough research or understanding to allow it to be considered seriously. To make matters worse the equal loudness curves, relating to the ear, vary with overall level so you'd need a kind of dynamic frequency shaping at the source that's based on output level at the ear. Complicated nonsense at this stage I feel. |
|
|
| danielwritesbac |
| quote: | Originally posted by ShinOBIWAN
I think he's talking about those fletcher/Munson equal loudness curves that show how the ear/brain interprets measured signals of equal loudness as being higher or lower in level depending on frequency. In a nutshell; we measure it flat but the ear/brain interprets this as non flat due to its own non linearity. |
YES!!! You can go measure ST Thompson/Bose amps and find it. You can also just look out the window and try to guess how many cars aren't equipped with bass-n-treble controls. Maybe just one--probably none. And, you can measure an uber-quality oil/wax audiophile cap. It does that warmth and air thing too.
And then you can decide if an amplifier should be "a device to please ears" or not. If ears are important, then maybe their response curve is also important.
Thanks ShinOBIWAN!! |
|
|
| ShinOBIWAN |
Erm don't thank me just yet.
In case you missed it I also wrote
Its all psychoacoustics with not enough research or understanding to allow it to be considered seriously. To make matters worse the equal loudness curves, relating to the ear, vary with overall level so you'd need a kind of dynamic frequency shaping at the source that's based on output level at the ear. Complicated nonsense at this stage I feel.
The people who research the field of psychoacoustics openly admit that their current findings and understanding of it are sketchy at best. To go ahead, such as you do, and broadly advocate such methods as being the true way despite nearly a century of working with the current standard of "keep the response flat" is very short sighted and idealistic. |
|
|
| BWRX |
| quote: | Originally posted by ShinOBIWAN
I think he's talking about those fletcher/Munson equal loudness curves that show how the ear/brain interprets a known and measured flat response as being higher or lower in level depending on frequency. |
I know what the Fletcher/Munson loudness curves are, but in what way are/should they be related to amplifiers? The purpose of an amplifier is to amplify a signal, not to alter it. To try and make an amplifier's response mimic the loudness curves is really useless because the final acoustic response of the complete system depends much more on the speakers and the room. |
|
|
| danielwritesbac |
| quote: | Originally posted by BWRX
Increasing the gain DOES NOT increase the power of the amplifier! |
False. Go try it. See how much more non-clipping audio is read by a prosound power meter.
I don't want to blow up your speakers, so do please have the DC protection cap present in the NFB. That's Ci on most of the relevant documents.
Resistors of 120k and 2.7k should achieve a gain setting of 45 at the NFB.
Unfortunately, you'll notice that the sonic signature of the amplifier has increased along with the power and dynamics.
I'm working on that coloration problem. |
|
|
| danielwritesbac |
| quote: | Originally posted by BWRX
I know what the Fletcher/Munson loudness curves are, but in what way are/should they related to amplifiers? The purpose of an amplifier to amplify a signal, not to alter it. To try and make an amplifier's response mmimic the loudness curves is really useless because the final acoustic response of the complete system depends much more on the speakers and the room. |
I'd sure like some help on that one. I don't have a clue whether its the sonic signature, interreactions with speakers, or whatever other errata is causing this particular lineup of amplifiers (National Semiconductor) to read out straight on test equipment and fail group testing (with humans as the test equipment).
Any ideas why the LM chips do this more than usual? |
|
|
| AndrewT |
| quote: | Originally posted by danielwritesbac
False. Go try it. See how much more non-clipping audio is read by a prosound power meter.
I don't want to blow up your speakers, so do please have the DC protection cap present in the NFB. That's Ci on most of the relevant documents.
Resistors of 120k and 2.7k should achieve a gain setting of 45 at the NFB.
Unfortunately, you'll notice that the sonic signature of the amplifier has increased along with the power and dynamics.
I'm working on that coloration problem. |
rubbish
| quote: | | very short sighted and idealistic |
more likely due to a lack of research leading to incorrect conclusions. |
|
|
| seventenths |
Daniel,
Not to try and lead you into a life of conformity, but for the purpose of this forum, adherence to the common vernacular would be appreciated.
7/10 (who measures audio performance in smiles) |
|
|
| BWRX |
| quote: | Originally posted by danielwritesbac
False. Go try it. See how much more non-clipping audio is read by a prosound power meter. |
Daniel, are aware what the term "gain" means?
Let's use an example. You have two ideal amplifiers operating off of a +/-10V supply with 4 ohm resistive loads. Amp A has a gain of 1V/V (0dB) and Amp B has a gain of 2V/V (6dB). Feed a sine wave with an amplitude of 1V to both amplifiers. Amp A will output a sine wave with an amplitude of 1V, while amp B will output a sine with an amplitude of 2V. Simple, right? Now increase the amplitude of the input signal to 5V. Amp A will output a sine wave with an amplitude of 5V, but Amp B will output a sine wave with an amplitude of 10V. If the input voltage is increased any further, the output of Amp B will be clipped. This means the peak, unclipped output power of Amp B is (10^2)/4 or 25W. Increase the amplitude of the input signal to 10V and Amp A will output a sine wave with an amplitude of 10V. If the input voltage is increased any further, the output of Amp A will be clipped. This means the peak, unclipped output power of Amp A is (10^2)/4 or 25W. A lower gain simply means that it takes a higher voltage signal to produce a higher level of output voltage. The output power of two identical amplifiers remains the same, independent of gain. |
|
|
| ShinOBIWAN |
| quote: | Originally posted by BWRX
I know what the Fletcher/Munson loudness curves are, but in what way are/should they related to amplifiers? The purpose of an amplifier to amplify a signal, not to alter it. To try and make an amplifier's response mmimic the loudness curves is really useless because the final acoustic response of the complete system depends much more on the speakers and the room. |
Sorry Brian, didn't mean to step on toes.
Whether the equal loudness curve is engineered into the recording, the amplifier or a corrective filter in the loudspeaker crossover its just not safe to go on an assumption that its anything like the right or correct way to do things because far more research is needed. That was the point I was trying to get across rather than exactly where it might be implemented. I did say "Complicated nonsense at this stage I feel" So I guess that sums up my feelings on the whole thing.
BTW Not to write off Psychoacoustics because it does have its merits and some aspects are well understood. The more advanced room correction methods are now able to take into account things like 1/6oct smoothing to better relate to our own hearing acuity or how large but very narrow Q peaks or dips in the FR can be left uncorrected because we can't hear them with music content. |
|
|
| BWRX |
| quote: | Originally posted by ShinOBIWAN
Sorry Brian, didn't mean to step on toes. |
No need to be sorry, your post about the loudness curves was quite correct. I was just stating their relevance to amplifiers. |
|
|
| ShinOBIWAN |
| quote: | Originally posted by BWRX
No need to be sorry, your post about the loudness curves was quite correct. I was just stating their relevance to amplifiers. |
Yes, although I didn't state it, I agree that an amplifier should be linear and if they ever do know enough about this to find merit in, and successfully implement, this then it should be done via signal shaping before the amp. |
|
|
| danielwritesbac |
| quote: | Originally posted by seventenths
Daniel,
Not to try and lead you into a life of conformity, but for the purpose of this forum, adherence to the common vernacular would be appreciated.
7/10 (who measures audio performance in smiles) |
Thank you! I am trying. ;) Opening day is the last day of April, and I have very little time to develop applied products.
That's just about as bizarre as a computer engineer (me) suddenly getting the assignment to lead a botanical research team. I'm a little bit lost, but getting "less lost" by the hour.
My first big contract is this coming Thursday, and thank goodness I have some products prepared.
Of course, I'd love to take the hotrod mini gainclone (LM1875) with me too, but its marvelous sound is highly reliant on a specific and outdated capacitor. So, I guess I'll just keep enjoying it in my own house. ;)
I would sincerely love to provide its sound as an *option* for others to enjoy, so I'm researching current production model parts to see if it can be duplicable.
Its the one that went "self propelled" this morning from too much audio. Andrew suggested using a larger transformer. . . as a weight. ;)
Anyway, its a great time to backstep a bit and say. . . While ears might not have solid relevance to amplifiers, there's a relevance to sales. I think that it must be a device to please ears, or else I would fail.
So, you might hear me mention a bit of market research factors in with the regular nonsense. ;) |
|
|
| danielwritesbac |
| quote: | Originally posted by ShinOBIWAN
Yes, although I didn't state it, I agree that an amplifier should be linear and if they ever do know enough about this to find merit in, and successfully implement, this then it should be done via signal shaping before the amp. |
YES!!! Score!!!
And, if you should choose to cause the amp to do that job itself?
ShinOBIWAN, thanks again!! You're the only one who understands what I'm talking about. Thanks so much for clarifying it. |
|
|
| Dr.EM |
Surely the point about equal loudness curves is not very relevant at playback stage since by nature it'll be engineered into the recording. The studio engineer monitors with his/her own ears so what you hear (from an accurate and flat playback system such as the studio system) should be the same output which sounded balanced to him/her. He/she might well turn down the violin or whatever in the mix because it sounds loud, even though in technical terms it isn't. Bass is nearly always the technically loudest aspect of a mix, it takes up the most space, because our ears are less sensitive to it. The point about nothing being right or wrong is very true though. Aiming for what sounds good to yourself is best.
Nice looking amp zdr, I like the use of the wood :) . If you do get problems with RF noise you could easily shield it internally with tin foil, metal mesh or copper clad boards. |
|
|
| danielwritesbac |
| quote: | Originally posted by Dr.EM
Surely the point about equal loudness curves is not very relevant at playback stage since by nature it'll be engineered into the recording. The studio engineer monitors with his/her own ears so what you hear (from an accurate and flat playback system such as the studio system) should be the same output which sounded balanced to him/her. He/she might well turn down the violin or whatever in the mix because it sounds loud, even though in technical terms it isn't. Bass is nearly always the technically loudest aspect of a mix, it takes up the most space, because our ears are less sensitive to it. The point about nothing being right or wrong is very true though. Aiming for what sounds good to yourself is best. |
It would be so nice if you'd take over the recording industry as soon as possible. ;) |
|
|
| ShinOBIWAN |
| quote: | Originally posted by danielwritesbac
YES!!! Score!!!
And, if you should choose to cause the amp to do that job itself?
ShinOBIWAN, thanks again!! You're the only one who understands what I'm talking about. Thanks so much for clarifying it. |
WHAT!!! Only you understand what your talking about, we're all here for the ride I feel. I'm amazed Daniel, how can one person take the meaning of something and completely and utterly misconstrue its true intention. That's a rare talent you have their please don't nurture it though.
<goes off to repeatedly bang head against wall> |
|
|
| MJL21193 |
| quote: | Originally posted by ShinOBIWAN
WHAT!!! Only you understand what your talking about, we're all here for the ride I feel. I'm amazed Daniel, how can one person take the meaning of something and completely and utterly misconstrue its true intention. That's a rare talent you have their please don't nurture it though.
<goes off to repeatedly bang head against wall> |
:D |
|
|
| ShinOBIWAN |
| quote: | Originally posted by Dr.EM
Surely the point about equal loudness curves is not very relevant at playback stage since by nature it'll be engineered into the recording. The studio engineer monitors with his/her own ears so what you hear (from an accurate and flat playback system such as the studio system) should be the same output which sounded balanced to him/her. He/she might well turn down the violin or whatever in the mix because it sounds loud, even though in technical terms it isn't. Bass is nearly always the technically loudest aspect of a mix, it takes up the most space, because our ears are less sensitive to it. The point about nothing being right or wrong is very true though. Aiming for what sounds good to yourself is best. |
Well that was one of those slap myself on the head moments. An excellent point and one I'd completely missed. |
|
|
| danielwritesbac |
| I just like your cat. ;) |
|
|
| danielwritesbac |
| I'd like to give for reference, the surround sound movement in addition to one other, yet unrelated, fac | | | |
