There is less "noise" if you ignore theirs and respond only to "signal" (however small it might be). Noise in the feedback loop doesn't help anything.
Filling half a page with pictures to show 1 recording, which is a vinyl transcription, possibly yours and therefore not available to anyone else to use as a reference is trolling of the highest order.
True. I cannot argue with you there. But still wish they would go away. I mean, what sane person walks into a mini show room and starts telling the salesman that unless they do a lap of a known circuit in less than n seconds they will buy a Peugeot?
I asked you to list the recordings you suggest others use as test albums, not show a screen shot of your rum removal or share copyrighted information via PM.
In case it isn't clear, by "list" I mean a text-only listing in the format Composer/Artist - Album Title - Track Number/Title (if applicable) - Record Label and Ordering Information (if applicable or necessary). One line per album. This way, readers will get the information they need quickly and concisely.
Differential input or not, I think you're missing the point that the MOD86 and PAR86 are a few orders of magnitude better than even a good standard LM3886 build. Towards the high end (>5 kHz), my implementation comes closer to being three orders of magnitude (1000x) better. That's just on the THD+N. Factor in the stellar PSRR and CMRR of the MOD/PAR86 and you have a winner.
Tom
Last edited:
Assuming you've already got solder, flux, and all necessary tools,
Heatsink thermal grease (Wakefield 120) is about USD 10.00 plus tax and shipping
If you're only going to build a single Power86 PCB, electronic components cost is about USD 24.00 to 26.00 at DigiKey or Mouser, plus tax and shipping.
So, PCBoard ($50) + parts + tax + shipping add up to something around USD 84.00 if you already have a supply of heatsink grease.
DigiKey and Mouser offer quite attractive volume discounts, so if you can assemble a buyer's consortium* for 25 boards, parts cost will fall rather dramatically. On the other hand, you'll pay for shipping twice: once from Mouser to (the consortium manager), and then again to each of the final buyers. Shipping boxes and envelopes are dirt cheap, check uline.com for the qty=25 pricing.
*Sometimes called a Group Buy here on diyAudio
Thanks, that's exactly what I needed to know. I have solder, thermal grease etc left over from an lm3875 I built years ago. Now my only problem is that I'm affraid if I repurpose the toroids from my old amp, they'll be too much for Mod 86.
What do you guys think, they're 330va 25+25. (In their present configuration with 8 diodes and 2x 10uF I get +/- 37v rails *edit and voltage drop when I connect amp it reads 35.8v)
Tom reccomends a 22+22 I think.
I plan to use either 1 Power86 with 6 Mod86 (but I doubt I can get away with that)
Or 2x 1 Power86 with 3 Mod86
Or 1 Power86 with 4 Mod86 (for mid/tweet) + 1 Power86 with 2 Mod86 (for bass)
I'm trying to run (tri-amp) a stereo pair of 3 ways (with active crossover after source before amp)
If I can get by with only 1 or 2 trafos and Power86 boards I might be able to squeeze it into the budget.
All comments appreciated.
AlexQS
What do you guys think, they're 330va 25+25. (In their present configuration with 8 diodes and 2x 10uF I get +/- 37v rails *edit and voltage drop when I connect amp it reads 35.8v)
Tom reccomends a 22+22 I think.
I plan to use either 1 Power86 with 6 Mod86 (but I doubt I can get away with that)
Or 2x 1 Power86 with 3 Mod86
Or 1 Power86 with 4 Mod86 (for mid/tweet) + 1 Power86 with 2 Mod86 (for bass)
I'm trying to run (tri-amp) a stereo pair of 3 ways (with active crossover after source before amp)
If I can get by with only 1 or 2 trafos and Power86 boards I might be able to squeeze it into the budget.
All comments appreciated.
AlexQS
Last edited:
I suggest you do some engineering calculations if you're planning to connect many amplifier boards to not-many PSU boards. Prove to yourself that you're not operating components beyond their datasheet spec limits. Quantify exactly how large or how small, your margin-of-safety really is.
You could start by looking at the LM3886 datasheet to estimate how many watts each amplifier PCB will draw from the power supply, when delivering output power to a loudspeaker load. (The number changes when the loudspeaker load changes from 4 ohms to 8 ohms). The images below might be helpful.
Then you could estimate the amount of power dissipated in the power supply board itself. There's a heatsink on the Power86 board; figure out whether it is necessary, or cosmetic.
Remember that the total power pulled out of the transformer secondary, is equal to the power dissipated in the PSU board, plus N times the amplifier board total power, where N is the number of amplifier boards per PSU board. How does this stack up against the transformer VA rating? Are you overloading the transformer? Have you left yourself plenty of safety margin?
Next estimate the voltage sag on the power supply when driving a heavy load. You know the PSU board's filter capacitance C, you know the load current I (it is simply N x loadpower_per_board / 35V), so you can calculate dV/dt. You know dt, it's 8.3 milliseconds (US) or 10 milliseconds (EU). So calculate the sag, dV. If the power supply voltage sags way WAY below 35V then the amplifier will clip prematurely (on supply troughs) and you don't want that.
Then I think it would be useful to read up about Ben Duncan's free power supply analysis software called PSUD2. Many people here on DIYA use it and swear by it. See whether PSUD2 helps you decide how many amplifier boards you can drive from one 330VA, 25V+25V transformer and one Power86 board.
If you don't know how to carry out this analysis and perform these calculations, that itself is probably a warning sign: Beware, you might be wading in over your head.
_
You could start by looking at the LM3886 datasheet to estimate how many watts each amplifier PCB will draw from the power supply, when delivering output power to a loudspeaker load. (The number changes when the loudspeaker load changes from 4 ohms to 8 ohms). The images below might be helpful.
Then you could estimate the amount of power dissipated in the power supply board itself. There's a heatsink on the Power86 board; figure out whether it is necessary, or cosmetic.
Remember that the total power pulled out of the transformer secondary, is equal to the power dissipated in the PSU board, plus N times the amplifier board total power, where N is the number of amplifier boards per PSU board. How does this stack up against the transformer VA rating? Are you overloading the transformer? Have you left yourself plenty of safety margin?
Next estimate the voltage sag on the power supply when driving a heavy load. You know the PSU board's filter capacitance C, you know the load current I (it is simply N x loadpower_per_board / 35V), so you can calculate dV/dt. You know dt, it's 8.3 milliseconds (US) or 10 milliseconds (EU). So calculate the sag, dV. If the power supply voltage sags way WAY below 35V then the amplifier will clip prematurely (on supply troughs) and you don't want that.
Then I think it would be useful to read up about Ben Duncan's free power supply analysis software called PSUD2. Many people here on DIYA use it and swear by it. See whether PSUD2 helps you decide how many amplifier boards you can drive from one 330VA, 25V+25V transformer and one Power86 board.
If you don't know how to carry out this analysis and perform these calculations, that itself is probably a warning sign: Beware, you might be wading in over your head.
_
Attachments
If the toroid and heat sinks worked with any LM38xx series IC, they'll be fine with the MOD86.
2x22 VAC gives about 28 V rectified when accounting for the current draw of the amp under modest load. With the Antek AN/AS-2222 I typically see a tad above 30 V at idle for nominal mains voltage.
The LM3886 is guaranteed (tested) to be able to deliver 7 A minimum of output current. That means the highest rail voltage that's guaranteed to work is 28 V for a 4 Ω load.
I also prefer to operate the ICs at no higher than 40 W of dissipation as this makes the thermals more manageable. You'll find this recommendation in the BPA-200 app note as well.
Those are the two main considerations that lead me to my recommendation of ±28 V rails for the LM3886 if used with 4 Ω loads. For 8 Ω*operation, you can crank the voltage up to ±35 V.
I go into quite a bit of detail on the thermal math as well as the power supply math on my Taming the LM3886 page.
You'll probably have a bit more ripple voltage than you'd like.
In that case, I'd recommend one Power-86 board per side (= woofer + mid + tweeter amp).
As to your question about BOM cost. Here are the numbers from Mouser as of today:
Modulus-86: $43.99
Power-86: $30.81
Parallel-86: $61.07
Tom
Last edited:
And if they're driving 4 ohm loads, attach a thermal cutout to the power transformer, because you'll exceed its 330 VA rating if you turn the volume all the way up.
For sine wave operation, you are correct. However, most of us listen to music rather than sine waves.
For music, the power draw depends on the crest factor of the music. Assuming a 14 dB crest factor, each channel would require 87 VA. Also, for a 3-way, this would mean delivering full power to all speaker drivers, continuously. This seems like a rather unlikely scenario.
If you're curious about the math behind this, I suggest checking out my Taming the LM3886 pages.
Tom
I agree with all the above.
A +-35Vdc supply at quiescent demand will sag significantly when asked to deliver full power.
Expect at least a 3Vdrop from quiescent voltage. i.e. +-35Vdc drops to less than +-32Vdc
Using the 8ohm plot in post1771, you will see that the lm3886 can deliver ~72W into 8r0 and dissipate ~29W (well below Tom's <40W recommendation) while this power is being delivered. From my own experiments and measurements I find much less than 68W from +-35Vdc and this has led me to using even higher quiescent voltage, I have gone as high as +-39Vdc and found no problem provided a decent size heatsink is used.
Only bass duty will stress the amplifier to near continuous power delivery, mid and treble duties are transients that last for very short durations, but repeat often.
The heatsink demands for mid and treble can be pretty low, because the duty cycle is low. Look at the dissipation when 1W is being delivered. 1W delivered into mid and into treble is very loud, but dissipates ~5W into the heatsink and the two amplifiers draw a total of 1+5+1+5 =12W. The bass amp could be 2W to 3W of average delivery and ~10W of dissipation.
The total average draw from the PSU for a 3way active could therefore be around 25W, for three channels rated @ 65W+65W+65W into 8ohms when supplied with +-35Vdc
The average dissipation load for the combined heatsink will be around 20W. A 0.5C/W sink will rise by ~15Cdegrees and will feel quite warm.
This shows that the first sentence made by Mark
is vitally important. Amplifier design is not guessing.
Last edited: