A "Wrench" has been thrown in the works.
I visited my parents this weekend and my dad gave me a pile of stuff that I had left there for years. There were two highend Alpine car subwoofers. 10"/4 ohm. They have excellent characteristics for the subs I want. I had forgotten about this stuff at least a dozen years ago. These are brand new never used because the car I intended to put them in was broken into and the rest of the system stolen.
I will need to adjust the design to accommodate the 4 ohm woofers. I think I will just need to lower the supply voltage.
I visited my parents this weekend and my dad gave me a pile of stuff that I had left there for years. There were two highend Alpine car subwoofers. 10"/4 ohm. They have excellent characteristics for the subs I want. I had forgotten about this stuff at least a dozen years ago. These are brand new never used because the car I intended to put them in was broken into and the rest of the system stolen.
I will need to adjust the design to accommodate the 4 ohm woofers. I think I will just need to lower the supply voltage.
What's the DCR?
29v rails? Your amp will probably handle those just as it is now. I'm assuming 1 amplifier per each woofer?
P.S. I'm using almost the same rail voltage (+/- 1volt same as yours) and my woofers are 4 ohms.
I used 4 ohms because an inductor of 3.5mh or larger allows for caps of 100uF or more, which makes for some very nice electronic dampening for the woofer. No "hand over mouth" effect from the woofer. P.P.S. Additional design adjustment: Small U-Channel aluminum stock, available at most hardware stores, over the "face" of LM3886TF, providing additional heatsinking and a much more secure mount. Thermally couple to LM3886TF with arctic silver paste (gray TT paste from Radio Shack local availability) to complete a plastic-to-metal thermal interface.
Last edited:
Daniel,
For sake of longevity of the chip I plan to use 22V rails. Antek's 18V 300VA transformers. One Transformer per channel.
As far as the chip itself is concerned I have heard the noninsulated "T" version sounds better? I am not sure why people are afraid of having the heatsink at 0V?? I would think that metal to metal contact would conduct the most heat from the chip?
I have heard of people using PC CPU fans to cool the chips. I think it would be really "cool" (pun intended) to use glycol/water coolant blocks on the chips and a remote mounted radiator and fan located where the noise of the fan would be less obtrusive. (thats something to think about down the road)
Anyway, The long and short of it is that the Antek Transformers should be able to give me 22 +/- rails all the way up to 10 Amps (5A on each rail peaks)
I have a torroid that I was hanging onto for a project that can power the tubes with plenty of room to spare. So I think the plan will be to "integrate" the preamp/Low pass crossover into the amp itself.
I will have plenty of power supply to spare so I can always look at the "Bridged" setup using a split load inverter tube circuit later.
I will update my schematic soon. By putting the crossover components between a gain and follower stage on the amp chassis itself it allows me to get one more piece of equipment out of the overall system and achieve the same ends.
My two way crossover for the mids and tweeters can easily be reworked since I have not built it yet. (the PC board is just collecting dust on a shelf) I can put a 6DB passive highpass at the input and the same on the mid amp to get the same results.
For sake of longevity of the chip I plan to use 22V rails. Antek's 18V 300VA transformers. One Transformer per channel.
As far as the chip itself is concerned I have heard the noninsulated "T" version sounds better? I am not sure why people are afraid of having the heatsink at 0V?? I would think that metal to metal contact would conduct the most heat from the chip?
I have heard of people using PC CPU fans to cool the chips. I think it would be really "cool" (pun intended) to use glycol/water coolant blocks on the chips and a remote mounted radiator and fan located where the noise of the fan would be less obtrusive. (thats something to think about down the road)
Anyway, The long and short of it is that the Antek Transformers should be able to give me 22 +/- rails all the way up to 10 Amps (5A on each rail peaks)
I have a torroid that I was hanging onto for a project that can power the tubes with plenty of room to spare. So I think the plan will be to "integrate" the preamp/Low pass crossover into the amp itself.
I will have plenty of power supply to spare so I can always look at the "Bridged" setup using a split load inverter tube circuit later.
I will update my schematic soon. By putting the crossover components between a gain and follower stage on the amp chassis itself it allows me to get one more piece of equipment out of the overall system and achieve the same ends.
My two way crossover for the mids and tweeters can easily be reworked since I have not built it yet. (the PC board is just collecting dust on a shelf) I can put a 6DB passive highpass at the input and the same on the mid amp to get the same results.
Local availability of very small aluminum U-channel is Lowes, Home Depot, the hardware store, etc. . .
This provides a slight pressure, which makes the thermal interface on the back of the chip thinner and more effective. The U-channel itself also adds to the heatsinking as well. This doesn't require a crushing pressure, but really just enough to squeeze the thermal paste.
This adaption helps LM3886 with 4 ohm woofers. Of course a more effective thermal interface is for causing the heatsink to be about the same temperature as the chip. The heatsink is hotter but the chip is cooler.
Here it is shown although mounting screws aren't added yet:
EDIT: needs screws at both sides of the chip to prevent accidental chip spin.
This provides a slight pressure, which makes the thermal interface on the back of the chip thinner and more effective. The U-channel itself also adds to the heatsinking as well. This doesn't require a crushing pressure, but really just enough to squeeze the thermal paste.
This adaption helps LM3886 with 4 ohm woofers. Of course a more effective thermal interface is for causing the heatsink to be about the same temperature as the chip. The heatsink is hotter but the chip is cooler.
Here it is shown although mounting screws aren't added yet:
EDIT: needs screws at both sides of the chip to prevent accidental chip spin.
Attachments
Last edited:
18V? Allied electric store brand 36vct, 4a ei core is Sku# 227-2065. MUCH more economical!
22vdc rails = endless clipping. Minimum 25vdc rails for reasonable results. Do remember that its solid state and also has the accursed SPIKE limiter circuit aboard the chip. The cure is voltage for headroom and decent heatsinking to avoid these problems.
If you're going to go for clipping, may as well avoid overspending on the transformer.
because its not 0v. Split-rail power supply will have one of the rails conduct directly to the tab of LM3886T. aka. . .ZAP!
Please don't die.
Venturi vent is also workable and makes no noises. Head spreader bars are another noiseless option (and available at the hardware store). Also, power supply design has a lot of options for reducing temperature.
You can cut temperature down with just the power supply board itself. First step would be to get rid of those MUR860's in trade for MR756 or ordinary diodes with miniature polyester caps as helpers. The "V+ to V-" cap can also drop temperature dramatically. Less amplification of noise makes nearly cold running chipamp. Then you can parallel it.
Last edited:
Rod Elliot: "Provided the amp is never allowed to go anywhere near the protection limits it sounds very good indeed. This is the rub - because of the comprehensive overload protection (which I have never liked in any form) this amp provides more and nastier artefacts as it clips than a "normal" amplifier."
Spike effectively cuts the headroom.
Avoiding that problem takes voltage.
It is listed on the bottom section of this page as is the quote:
Single Chip 50W Stereo Amplifier
In comparison to a vacuum tube, the chip can output only a fraction as much music on the same wattage rated amplifier.
Remember that simple Stancor EI core 20-0-20 posted earlier? 20x1.43=28.6vdc for each rail; straight on spec for using 4 ohm speakers. Its Allied SKU# 928-8566
Spike effectively cuts the headroom.
Avoiding that problem takes voltage.
It is listed on the bottom section of this page as is the quote:
Single Chip 50W Stereo Amplifier
In comparison to a vacuum tube, the chip can output only a fraction as much music on the same wattage rated amplifier.
Remember that simple Stancor EI core 20-0-20 posted earlier? 20x1.43=28.6vdc for each rail; straight on spec for using 4 ohm speakers. Its Allied SKU# 928-8566
Last edited:
You are priceless Dan!
This world would be so much more boring without you in it.
You are priceless Dan!
This world would be so much more boring without you in it.
Thank you!
You know that point right before a regular amplifier clips? Well, by then the LM3875/LM3886 have already engaged the spike system, lending a "woman trying to outscream her burglar alarm" coloration that's just unseemly. Rod Elliot said there were caveats.
The original poster has much vacuum tube information, and along with that may come assumptions with voltage causing amplification and/or expected usable power output. However, voltage will vary LM3886 only between clipping and overheating (not in expected proportion); whilst usable power output is only a fraction.
LM3886 as a subwoofer amplifier on 22vdc rails? That's just poor. With such pitifully low voltage, I'd rather use LM1875's or any other amp/driver multipurpose chip's much cleaner output, whereby you can, at least, make use of its full potential. But, what I'd like to do is make use of LM3886 and that takes voltage. I was trying to illustrate this. . . and continue to fail.
For 4 ohm speakers, one should avoid crippling LM3886 leaving it screaming for help; but, instead one should use LM3886 paralleled, turning it into a more seemly amplifier. Unfortunately, I've failed to illustrate that as well.
How to better convey this? Um. . . help!!
Last edited:
A 60vct free transformer? Perhaps. By cutting the center link, you have a 30vac + 30vac transformer. Two of "big silver shiny square" 20+ amper bridge rectifiers plus eight of 10nf polyester capacitors (rectifier and caps are in the drawer at the Radio Shack) for a pair of "snubbed rectifiers" (as seen in radio) will significantly sink the voltage output of the power supply. Its approximately 35vdc+35vdc dual-rail supply that way. This can do a good job with Parallel LM3886 into 4 ohm speakers. It will perform well enough for full bandwidth hi-fi use.
In this example, the probability of pretty bass is high, but the dynamic punch will need to come from the preamp.
Mark Houston demonstrates snubbing the rectifier here: Synergy - LM3875 Gainclone Amplifier (Chip Amp)
Cutting the centerlink out of the free/salvage 60vct to make a dual 30,30 and then using two of Mark's snubbed rectifiers along with a Parallel LM3886 is workable indeed. That's a nice 35 watts of super-clean audio plus some extra capacity for dynamics.
In this example, the probability of pretty bass is high, but the dynamic punch will need to come from the preamp.
Mark Houston demonstrates snubbing the rectifier here: Synergy - LM3875 Gainclone Amplifier (Chip Amp)
Cutting the centerlink out of the free/salvage 60vct to make a dual 30,30 and then using two of Mark's snubbed rectifiers along with a Parallel LM3886 is workable indeed.
That's a nice 35 watts of super-clean audio plus some extra capacity for dynamics.
Last edited:
OK,
Here is where everyone confuses me again.
In tube amplification (except for some obscure designs) amplification of signal is mostly done with very low current and voltage amplification.
BUT, in the power section using transformers we Oscillate BOTH the current AND the voltage to "transform" the relatively high voltage and low current to low voltage and Higher current (relative to each other) using Watts Law we get the POWER into the load.
Heres where I get confused. In this chip amp thing we need extra voltage for headroom? That excess voltage has to go somewhere (HEAT). We also have peak current demands of Sqrt((2XPo)/RL)
So assuming a 50 Watt output I need
Peak voltage of sqrt(2xRLxPo) = 14V
and Peak Current of Sqrt((2XPo)/RL) = 5Amps
Both of those numbers need to be DOUBLED for +/- supplies so I need 280Watts for 50 Watts output?
WHY?
The supplies are directly coupled to the load. So are you telling me that 230 Watts of power is dissipated in the chip?
Here is where everyone confuses me again.
In tube amplification (except for some obscure designs) amplification of signal is mostly done with very low current and voltage amplification.
BUT, in the power section using transformers we Oscillate BOTH the current AND the voltage to "transform" the relatively high voltage and low current to low voltage and Higher current (relative to each other) using Watts Law we get the POWER into the load.
Heres where I get confused. In this chip amp thing we need extra voltage for headroom? That excess voltage has to go somewhere (HEAT). We also have peak current demands of Sqrt((2XPo)/RL)
So assuming a 50 Watt output I need
Peak voltage of sqrt(2xRLxPo) = 14V
and Peak Current of Sqrt((2XPo)/RL) = 5Amps
Both of those numbers need to be DOUBLED for +/- supplies so I need 280Watts for 50 Watts output?
WHY?
The supplies are directly coupled to the load. So are you telling me that 230 Watts of power is dissipated in the chip?
I do not think that the power is dissipated in the chip it is not used. So why such a high VA rated transformer is needed?
Assuming I have an unlimited budget I could build a 100Watt Tube amp with a Power Transformer of only 250VA. 500V 500mA
I am sure I am looking at this in an oversimplified way and also being somewhat "Naive" to solid state designs.
Maybe someone can "straigten" me out.
Assuming I have an unlimited budget I could build a 100Watt Tube amp with a Power Transformer of only 250VA. 500V 500mA
I am sure I am looking at this in an oversimplified way and also being somewhat "Naive" to solid state designs.
Maybe someone can "straigten" me out.
14Vpk and 5Apk indicate a resistance of 2r8. Why?
Double up the supplies to give 28Vpk and 10Apk, no!.
Doubling the supplies (for dual polarity) gives 14Vpk but the 5Apk remains the same.
However that peak voltage only drives the peak current for an instant during each half cycle. During the rest of the time the currents are less, twice they are zeroApk. Does zero amperes equate to an effective zero Watts. No, because again that is only for an instant.
One needs to look at averages over the whole waveform cycle.
Here we use Vrms and Arms to determine the average power.
14Vpk = 9.9Vrms, The rms current through a 4ohm load will be ~2.5A.
The average power is ~25W, not 280W.
Now this is impossible 25W input for 50W output.
I think there's something wrong with your numbers.
Look back at 14Vpk. It has all gone wrong at your first bit of arithmetic.
Although I'm not currently capable of the maths for that, instead, I can give you a point of comparison.
amperage
A capactive multiplier or regulator will break the Linear relationship between big [amperage] transformer for big woofer or smaller [amperage] transformer for small woofer. Within reason, the number of chipamps running on a given transformer doesn't change the bass output at end result. Its merely a linear relationship to the bass frequencies supported no matter if you plan to make use of the full amperage or not.
voltage
And now back to Rod Elliot on the topic:
Power Amplifier Clipping
"Sound Impairment Monitor (SIM) - Is This The Answer?"
One of those Rod Elliot articles contains "the big difference" between solid state and vacuum tube.
I apologise that I couldn't illustrate it for you.
National Semiconductor's spike system approximately doubles the problem clipping of solid state as well as the pre-clipping stage, which is a form of compression. Voltage is the cure--a much more powerful amp than you intend to use is how you "size" solid state. Spike chips don't make racket simultaneously during the pre-clip stage and so many people choose parallel operation for cancellations on a portion of the noises.
Therefore, you may wish to consider parallel operation.
On topic of the spike system, this is how LM1875 beats LM3875 on useful power output. This is also how the 32 watts of STK465 0.08%thd at 32w (or a similar discrete amp) easily beats LM3886 on useful power output. At the speaker's output, 3db differences with double the power. . . needs to be a useful 3db, and that takes voltage.
A useful 3db? Its just like this:
Vacuum tube baseline: double the power to use solid state instead.
Solid state baseline: double the power to use NatSemi spike chip instead.
Fortunately providence has sent us 4 ohm woofers. Now we need a parallel lm3886 on appropriate voltage.
The results should be lovely.
Ah, please don't forget the spike circuit--25 watts of "intact" audio signal might be optimistic.
Apparently (judging from some of the other responses) you need some competent advice.
As AndrewT says, power output calculations are wrought from rms values of volts and amperes, as we are dealing with AC, for example:
For that 50 watts output into a 4 ohm load the rms voltage would be 14.14 at 3.53 amps rms. In terms of peak voltage output this is 20 volts, or 40 volts peak to peak.
The supply voltage would need to be higher than the peak voltage to account for losses, so lets say 25VDC positive and 25VDC negative or in other words, +/-25VDC. Now, the VA rating for the transformer is easy enough to determine - you know you need 3.53A of current at 50 watts output and since your rail voltage of +/-25VDC you need a 18-0-18 VAC transformer. My quick and painless way of calculating VA is to multiply half the AC voltage (in this case 18VAC) by the rms current (in this case 3.53) and this give ~60VA which would be sufficient but bumping it up to 100VA (at the same voltage) will not hurt, giving some current headroom and larger transformers operate more efficiently.
Hope this helps.
MJL, Andrew,
THANK YOU. I understand it now!!!
Again, most of my confusion comes from working with SINGLE supply circuits for almost all of my experience. I did do the math wrong. First mistake was the square root of 400 is 20 not 14 (not sure where I got that?)
Second mistake was doubling the voltage AND the current.
Correcting those to
20V peak +/- and Peak Current of 5A
Need to add the dropout voltage of the chip 4V plus regulation and high/low line conditions I need about 28V +/- with Peak of 5A.
Assuming it is a BASS amp then I really do not want to "scrimp" on the Amperage side correct? Unless I want about a FARAD of capacitance.
So a transformer per channel of 18-20V @ 5Amp or 100VA is suitable?
THANK YOU. I understand it now!!!
Again, most of my confusion comes from working with SINGLE supply circuits for almost all of my experience. I did do the math wrong. First mistake was the square root of 400 is 20 not 14 (not sure where I got that?)
Second mistake was doubling the voltage AND the current.
Correcting those to
20V peak +/- and Peak Current of 5A
Need to add the dropout voltage of the chip 4V plus regulation and high/low line conditions I need about 28V +/- with Peak of 5A.
Assuming it is a BASS amp then I really do not want to "scrimp" on the Amperage side correct? Unless I want about a FARAD of capacitance.
So a transformer per channel of 18-20V @ 5Amp or 100VA is suitable?
If I went with two of these Antek 20V 200VA I think I would be pretty safe.
I have a similar sized (dimensionally) Torroid with taps suitable for the Tube section.
This would give me some symmetry on the chassis, and when I have the funds I could purchase steel shells to even further the look.
I have a similar sized (dimensionally) Torroid with taps suitable for the Tube section.
This would give me some symmetry on the chassis, and when I have the funds I could purchase steel shells to even further the look.
I'm using a 20,0,20 vac 4 amper center tap to provide the minimum voltage for successful playback. . . without the Spike protection system sounding off.
Your transformer, a 20+20, http://www.antekinc.com/pdf/AN-2220.pdf can do exactly the same and you also have the option of dual rectifiers.
However, in the case of dual rectifiers (dual rectifiers make for slightly less voltage), the following transformer may make for more seemly results: Antek - AN-2222 It is the 22+22 version, because the extra couple of volts will get you farther away from the Spike system noise (and the extra couple of volts will do absolutely nothing else).
These are all workable transformer choices. Kudos.
Your transformer, a 20+20, http://www.antekinc.com/pdf/AN-2220.pdf can do exactly the same and you also have the option of dual rectifiers.
However, in the case of dual rectifiers (dual rectifiers make for slightly less voltage), the following transformer may make for more seemly results: Antek - AN-2222 It is the 22+22 version, because the extra couple of volts will get you farther away from the Spike system noise (and the extra couple of volts will do absolutely nothing else).
These are all workable transformer choices. Kudos.
YES!!!
28.6vdc per each rail, with over 4a capacity. Oh yes!! This works practically too! P.S. Dual bridge rectifier (8 diode), multiply ac by 1.25 (using 22+22 transfo). . . or for solo bridge rectifier (4 diode), multiply ac by 1.43 (using 20-0-20 transfo). Power supply caps will charge up to meet the voltage peak.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.