There are certainly times when using different models of amp for the woofer, tweeter or mid is desirable. However, do certain models exhibit different phase shifts? By looking at the circuits how would you figure out the overall phase shift of an amplifier? For instance I may want to use:
http://sound.westhost.com/project19.htm
For the tweeter and:
http://sound.westhost.com/project101.htm
For the mid+bass. Would that potentially cause issue, or is the phase shift at these frequencies only ever going to be 180, 0 or 360? As far as I know it is only capacitive coupling that can cause a phase shift and usually this is only at much lower frequencies with the size of coupling capacitor concerned.
http://sound.westhost.com/project19.htm
For the tweeter and:
http://sound.westhost.com/project101.htm
For the mid+bass. Would that potentially cause issue, or is the phase shift at these frequencies only ever going to be 180, 0 or 360? As far as I know it is only capacitive coupling that can cause a phase shift and usually this is only at much lower frequencies with the size of coupling capacitor concerned.
Hi,
arrange for both amplifiers to be either both non-inverting or both inverting (not recommended). That gets rid of the phase inversion problem.
There is a small amount of phase shift in all amplifiers. But it usually only becomes serious as the frequency approaches the F-3db cut-off points. We can usually ignore this.
Check the phase coming from the filters. They are usually non-inverting, although a few topologies are inverting.
Now the big Q.
What are the impedances and sensitivities of your drivers?
I am surprised at the difference in voltage drive you are planning to build into the two amplifiers.
50W into 8r0 is 28.3Vpk.
180W into 8r0 is 53.7Vpk.
That is quite a difference in voltage, 5.6dbV!
To clarify why I'm prompting you to think about sensitivity and voltage drive, I believe that the peak SPL from all the speakers in a music playing system (and this applies to film sound tracks as well) should be similar.
If the drivers all have the same sensitivity and the same impedance, then they should each be capable of being driven to the same peak voltage.
If the sensitivities are different then this must be taken into account when selecting the power amplifier voltage output.
If the impedances are different then the effective load determines how robust the output stages need to be (to be reliable).
Forget all those do-gooders that talk enormous power for bass. The best that can achieve is disco sound, unless that's your target.
arrange for both amplifiers to be either both non-inverting or both inverting (not recommended). That gets rid of the phase inversion problem.
There is a small amount of phase shift in all amplifiers. But it usually only becomes serious as the frequency approaches the F-3db cut-off points. We can usually ignore this.
Check the phase coming from the filters. They are usually non-inverting, although a few topologies are inverting.
Now the big Q.
What are the impedances and sensitivities of your drivers?
I am surprised at the difference in voltage drive you are planning to build into the two amplifiers.
50W into 8r0 is 28.3Vpk.
180W into 8r0 is 53.7Vpk.
That is quite a difference in voltage, 5.6dbV!
To clarify why I'm prompting you to think about sensitivity and voltage drive, I believe that the peak SPL from all the speakers in a music playing system (and this applies to film sound tracks as well) should be similar.
If the drivers all have the same sensitivity and the same impedance, then they should each be capable of being driven to the same peak voltage.
If the sensitivities are different then this must be taken into account when selecting the power amplifier voltage output.
If the impedances are different then the effective load determines how robust the output stages need to be (to be reliable).
Forget all those do-gooders that talk enormous power for bass. The best that can achieve is disco sound, unless that's your target.
Thanks for your helpful answer 
If I ever did do this I planned to use an XG18 mid and MDT33 tweeter with 86 and 92.5 db sensitivities respectively. With both at 8 ohm the tweeter theoretically only requires roughly one quarter of the power to produce the same output (I think). Rod himself also noted that average tweeter power was much lower than the mid or bass with normal content, although it clipping would be pretty awful so the headroom is worthwhile.
I should have mentioned I planned on using the P101 at a lower voltage to produce around 100w for the mid and at around +/-63v for the bass unit; although this is a more sensitive driver, (Monacor SPH-275C) at 90db so theoretically requires less power than the mid I have heard that bass (sub 250hz content) recorded in music and film soundtracks is around 10db louder than the rest in peak output. This is also something I've noticed when producing my own recordings, although the "trend" now is often to limit everything totally flat
.
The filter would be the Linkwitz Riley 24db topology so phase shifts at filter level should not be a concern.
If I ever did do this I planned to use an XG18 mid and MDT33 tweeter with 86 and 92.5 db sensitivities respectively. With both at 8 ohm the tweeter theoretically only requires roughly one quarter of the power to produce the same output (I think). Rod himself also noted that average tweeter power was much lower than the mid or bass with normal content, although it clipping would be pretty awful so the headroom is worthwhile.
I should have mentioned I planned on using the P101 at a lower voltage to produce around 100w for the mid and at around +/-63v for the bass unit; although this is a more sensitive driver, (Monacor SPH-275C) at 90db so theoretically requires less power than the mid I have heard that bass (sub 250hz content) recorded in music and film soundtracks is around 10db louder than the rest in peak output. This is also something I've noticed when producing my own recordings, although the "trend" now is often to limit everything totally flat
. The filter would be the Linkwitz Riley 24db topology so phase shifts at filter level should not be a concern.
Hi Dr.EM,
Normally one would use at least the same make of amp, just different power ratings. I don't think it's that important to most people on average. The important thing is that you have a way to measure the output of the amp to be able to match sensitivities of the amps, or correct for your speaker mismatch there.
If one amplifier inverts and the other doesn't, who cares? What you then do is reverse your speaker polarity on that amp. All fixed.
One more point. Do not attempt to use the power meters on the amplifiers. They are not normally accurate, not even close.
-Chris
Normally one would use at least the same make of amp, just different power ratings. I don't think it's that important to most people on average. The important thing is that you have a way to measure the output of the amp to be able to match sensitivities of the amps, or correct for your speaker mismatch there.
If one amplifier inverts and the other doesn't, who cares? What you then do is reverse your speaker polarity on that amp. All fixed.
One more point. Do not attempt to use the power meters on the amplifiers. They are not normally accurate, not even close.
-Chris
It's a really good idea to use different models of amps to drive the different speakers in a biamp setup. Biamping allows you to tailor the performance of the amplifier to each driver, rather than having a compromise.
For example, you could design your tweeter amp for lower power, but faster slew rate, than your woofer amp.
Phase shifts should be negligible for most any audio amp in the audio band, so don't worry about that.
I'm currently building an amp to biamp my monitor speakers in my study. I'm using a home-built 50W mosfet amp based on the AEM6000 for the mid-woofer (a Vifa P13WH), and an amp using eight paralleled TPA6120 headphone driver ICs for the Vifa D27TG-35. The tweeter amp is only good for 10W or so, but this application uses a crossover frequency of 3300Hz (or thereabouts) so that's oodles.
The nice thing about the TPA6120 is that it's got a really stupendous slew rate. It'll do a 10V step in 5ns, so its intermods are practically non-existant, and it's got gain out past 100MHz. With the exception of being designed for 64 Ohm loads (and paralleling a bunch fixes that) it's the perfect tweeter amp.
One thing to be careful of is that the levels are set correctly. My tweeters are 6 Ohm, with 91dB sensitivity. My mid-woofers are 8 Ohm, with 88dB sensitivity. Taking the different impedences into account, and using my mid-woofer amp gain of 16 as the reference, I calculated that I'd need to set the tweeter amp gain to 9.8.
Then if I put 1V RMS in, I get 9.8V at the tweeter, and 16V at the mid-woofer. That equates to 16W into the tweeter, and 32W into the woofer, a difference of 3dB, so after their difference in sensitivity is adding in, it should be nicely flat.
For example, you could design your tweeter amp for lower power, but faster slew rate, than your woofer amp.
Phase shifts should be negligible for most any audio amp in the audio band, so don't worry about that.
I'm currently building an amp to biamp my monitor speakers in my study. I'm using a home-built 50W mosfet amp based on the AEM6000 for the mid-woofer (a Vifa P13WH), and an amp using eight paralleled TPA6120 headphone driver ICs for the Vifa D27TG-35. The tweeter amp is only good for 10W or so, but this application uses a crossover frequency of 3300Hz (or thereabouts) so that's oodles.
The nice thing about the TPA6120 is that it's got a really stupendous slew rate. It'll do a 10V step in 5ns, so its intermods are practically non-existant, and it's got gain out past 100MHz. With the exception of being designed for 64 Ohm loads (and paralleling a bunch fixes that) it's the perfect tweeter amp.
One thing to be careful of is that the levels are set correctly. My tweeters are 6 Ohm, with 91dB sensitivity. My mid-woofers are 8 Ohm, with 88dB sensitivity. Taking the different impedences into account, and using my mid-woofer amp gain of 16 as the reference, I calculated that I'd need to set the tweeter amp gain to 9.8.
Then if I put 1V RMS in, I get 9.8V at the tweeter, and 16V at the mid-woofer. That equates to 16W into the tweeter, and 32W into the woofer, a difference of 3dB, so after their difference in sensitivity is adding in, it should be nicely flat.
I had considered the different amps strengths and weaknesses to some extent but perhaps not as fully as I could have.
The P101 is perfect for bass with its very high damping factor and high current delivery, i'm using it in a sub with good results. I hadn't put much consideration into its use for mids beyond that it would be easy to make a second similar amplifier unit. By all reports its a good amp accross the board and has a decent output to get that low sensitivity mid going.
The Gainclone style amp specs with lower noise which I felt may be an advantage with the tweeter being most sensitive and also the typical "hissing" noise being most prominent at tweeter frequencies. Compared to your TPA6120 its slew rate isn't so impressive but looks pretty fast in general terms. I take it 50W is possibly excessive at above 2khz or so? The MDT33 supposedly handles 200W, thats an awful lot of treble power it seems
For calibrating I was actually considering mics. Presumably pointing a flat response mic at each driver while it is playing a tone in the middle of its range and adjusting each ones level will achieve a fairly flat response? I suppose variations in thier frequency response may upset this though. Either way, final adjustment will likely be made by ear.
I own a basic oscilloscope so I can check things like phase, amplitude, oscillations, power output etc on that to confirm everything is at it should be. Built in 1984, bought for £30 second hand; works fine though
The P101 is perfect for bass with its very high damping factor and high current delivery, i'm using it in a sub with good results. I hadn't put much consideration into its use for mids beyond that it would be easy to make a second similar amplifier unit. By all reports its a good amp accross the board and has a decent output to get that low sensitivity mid going.
The Gainclone style amp specs with lower noise which I felt may be an advantage with the tweeter being most sensitive and also the typical "hissing" noise being most prominent at tweeter frequencies. Compared to your TPA6120 its slew rate isn't so impressive but looks pretty fast in general terms. I take it 50W is possibly excessive at above 2khz or so? The MDT33 supposedly handles 200W, thats an awful lot of treble power it seems
For calibrating I was actually considering mics. Presumably pointing a flat response mic at each driver while it is playing a tone in the middle of its range and adjusting each ones level will achieve a fairly flat response? I suppose variations in thier frequency response may upset this though. Either way, final adjustment will likely be made by ear.
I own a basic oscilloscope so I can check things like phase, amplitude, oscillations, power output etc on that to confirm everything is at it should be. Built in 1984, bought for £30 second hand; works fine though
It certainly seems unlikely. It's not a spec I chose it for, its just a well regerded tweeter with a smooth sound:
http://www.morelhifi.com/products/pdf/mdt 33.pdf
I don't like the overbright, exaggerated sibilence sound I seem to be getting woth my DM602's. I use the treble control to tame it but it's still excessive. I'm not sure how much of this is down to the way things are recorded now though but so long as the speaker handles it without getting all piercing/resonant sounding it should be ok.
http://www.morelhifi.com/products/pdf/mdt 33.pdf
I don't like the overbright, exaggerated sibilence sound I seem to be getting woth my DM602's. I use the treble control to tame it but it's still excessive. I'm not sure how much of this is down to the way things are recorded now though but so long as the speaker handles it without getting all piercing/resonant sounding it should be ok.
Its good news that you also think its the system responsible, otherwise i'd never be very satisfied with a lot of music. The system can always be changed however.
I would be putting this all together right now if it wasn't for one thing. That is the space I have available. There really isn't room in my bedroom for it and the small space will not help the sound at all either way. It would be possible to accomodate by changing a lot of whats in there and moving things around but it really isn't ideal
I would be putting this all together right now if it wasn't for one thing. That is the space I have available. There really isn't room in my bedroom for it and the small space will not help the sound at all either way. It would be possible to accomodate by changing a lot of whats in there and moving things around but it really isn't ideal
Hi Dr.EM,
It's amazing how the sound of an amplifier can affect the sound of a tweeter. I have not yet heard a chip amp that does well there. The always seem to have a "non-smooth", or strident sound. Many other types of amplifiers too. Component choice will affect details a fair amount.
Hi suzyj,
I have heard many complimentary differential pair amplifiers and preamps. They all share an "edgy" quality. Not something that stands out immediately (assuming a well designed and built amplifier). This is true whether the device uses J-FETs or BJT pairs. I have even hand matched the parts and while that improves things, that quality remains. A single long tailed pair doesn't seem to have this sound. J-FET or valve.
Any thoughts? I only brought this up because the designs you tend to use are complimentary differential. I was wondering if you had experimented with other topologies.
I have also noticed that BJT pairs sound better when driven with a low impedance, high impedance drive may result in a "grainy" quality.
Your point about matching the amplifier to the job at hand is well made.
-Chris
It's amazing how the sound of an amplifier can affect the sound of a tweeter. I have not yet heard a chip amp that does well there. The always seem to have a "non-smooth", or strident sound. Many other types of amplifiers too. Component choice will affect details a fair amount.
Hi suzyj,
I have heard many complimentary differential pair amplifiers and preamps. They all share an "edgy" quality. Not something that stands out immediately (assuming a well designed and built amplifier). This is true whether the device uses J-FETs or BJT pairs. I have even hand matched the parts and while that improves things, that quality remains. A single long tailed pair doesn't seem to have this sound. J-FET or valve.
Any thoughts? I only brought this up because the designs you tend to use are complimentary differential. I was wondering if you had experimented with other topologies.
I have also noticed that BJT pairs sound better when driven with a low impedance, high impedance drive may result in a "grainy" quality.
Your point about matching the amplifier to the job at hand is well made.
-Chris
Thanks for your input there, I certainly don't want to end up with any harshness in the treble again. I had considered class A, but Rod has no PCB for his design and thermal management is a bit of a task 
I was also thinking about soft start for the system. I have planned to have a seperate small unit for power distibution containing the soft start and transformer for the crossover/preamp. Ideally it seems good to sequentially start the amplifiers but this requires a lot of mains wiring which I try to keep to a minimum. I'm wondering now if one soft start could be used for the bunch so to speak. With 3 transformers (625va, 500va, 300va) equaling 1425va can one soft start be used for all? It seems around 20ohm is required and I plan to use 10 220ohm 50W units in parallel (to give safety for up to about 3 seconds, it's really not that expensive at £12 for all the resistors). Will everything work well like this or is there some factor I haven't considered?
I was also thinking about soft start for the system. I have planned to have a seperate small unit for power distibution containing the soft start and transformer for the crossover/preamp. Ideally it seems good to sequentially start the amplifiers but this requires a lot of mains wiring which I try to keep to a minimum. I'm wondering now if one soft start could be used for the bunch so to speak. With 3 transformers (625va, 500va, 300va) equaling 1425va can one soft start be used for all? It seems around 20ohm is required and I plan to use 10 220ohm 50W units in parallel (to give safety for up to about 3 seconds, it's really not that expensive at £12 for all the resistors). Will everything work well like this or is there some factor I haven't considered?
Thats good. Yeah, the preamp needs to be on before the amps and off after to avoid the thuds and pops, I found that out the hard way making my subwoofer, an unplanned muting circuit was added (actually based on the soft start design).
This would likely make a good tweeter amp:
http://sound.westhost.com/project83.htm
Although despite it looking simple (enough so I could build a PCB, or even point to point wire it, particularly the bottom variation with its associated preamp) it would cost rather a lot, prohibitively so I fear. With 4 large heatsinks, the larger case to house them, larger transformer and that IRF350 MOSFET costing nearly £15 a piece
Wish I could compare these designs properly. If I could have got to the meet I might have a better idea about how different designs actually sound.
This would likely make a good tweeter amp:
http://sound.westhost.com/project83.htm
Although despite it looking simple (enough so I could build a PCB, or even point to point wire it, particularly the bottom variation with its associated preamp) it would cost rather a lot, prohibitively so I fear. With 4 large heatsinks, the larger case to house them, larger transformer and that IRF350 MOSFET costing nearly £15 a piece
Wish I could compare these designs properly. If I could have got to the meet I might have a better idea about how different designs actually sound.
anatech said:
I like the complementary differential topology because when it's properly executed, it tends to cancel out many of the non-linearities in the VAS transistors. This allows me to use a little less emitter degeneration, and thus get more overall open-loop gain than I'd otherwise be able to get with a given number of stages, resulting in lower distortion.
It results in complicated amplifiers, certainly, and I've noted a lot of people on these boards are frightened by complexity.
I think the quality of execution counts for a hell of a lot more than the chosen topology. Especially when an amplifier is highly optimised, its performance can be completely destroyed by even small stuff-ups. A good case in point is my first go at a 50W MOSFET amp. I thought I had a good understanding of the distortion mechanisms, and designed a really neat amp. I laid it out well, built it with high quality components and a well made (and expensive!) double sided 4 ounce board, and on testing the result, and listening to it, found it truly sucked. It wasn't a wasted effort though, as it forced me to learn about thermal matching in the VAS stage (something that no number of simulations can help with), and that's a good thing. The next one will (hopefully!) be better as a result.
But something that it has taught me is that I don't have nearly as much control over the distortion cancellation as I thought I did - there are more variables at play there than I knew of, and I'm sure there are still more subtleties than I'm accounting for. It's a continual learning exercise, which I quite like.
So I'll happily admit that much of my attachment for complementary differential topologies is simple prejudice. One of these days I'm sure I'll get bored with this topology, and try others. For now though, my ears are liking what I'm doing.
Hi suzyj,
Fair enough. I've read some of your web page. Impressive. You have much more energy that I do these days. The amount of things you have finished (And finished well) is impressive.
As for layout and thermals, yes. Most people make errors there. I learned early on to match everything if possible and thermally couple complimentary parts. I've found the shortest signal path within reason generally is a good place to start. Like you, I design and build. I don't know how to simulate yet. I build two. One with matched parts and one with purposely mismatched parts. I like to see how critical the parts are in a design.
Some day I'd like to see you try a J-FET long tailed pair feeding another balanced BJT pair as a VAS. This seems to work well. Cascoding both pairs brings benefits (this you know already). Simple and not. You still get your push - pull VAS that you like.
Cheers!
Fair enough. I've read some of your web page. Impressive. You have much more energy that I do these days. The amount of things you have finished (And finished well) is impressive.
As for layout and thermals, yes. Most people make errors there. I learned early on to match everything if possible and thermally couple complimentary parts. I've found the shortest signal path within reason generally is a good place to start. Like you, I design and build. I don't know how to simulate yet. I build two. One with matched parts and one with purposely mismatched parts. I like to see how critical the parts are in a design.
Some day I'd like to see you try a J-FET long tailed pair feeding another balanced BJT pair as a VAS. This seems to work well. Cascoding both pairs brings benefits (this you know already). Simple and not. You still get your push - pull VAS that you like.
Cheers!
AndrewT said:
It was all fleshed out in http://www.diyaudio.com/forums/showthread.php?s=&threadid=105070, but the long and short of it is that differential pairs, wherever they are in the circuit, need to be tied together thermally, or else the advantages of using a differential pair go out the window.
suzyj said:
Sorry for being picky, but only if the dynamic current variation through the devices is excessively high (in terms of significant dynamic self heating) for their die size / dissipation capability.
No thermal coupling used here; just bog standard to-92's in a differential input and a differential folded cascode VAS:
http://www.hardwareanalysis.com/content/article/1842.4/extrema_reference_class_a_diy_amplifier/
http://www.hardwareanalysis.com/content/article/1842.5/extrema_reference_class_a_diy_amplifier/
The little 12W class A amp on my webpage does <0.001% THD at 20kHz with the 30MHz BJT's too, and no diff stage device thermal coupling.
Cheers,
Glen
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.