I saw a recommendation to set the NFB roll-off >half an octave below the Input Filter.
I found that worked for me and my amplifiers.
I have recommended that ever since.
i.e. use factor of >=sqrt(2).
I found that worked for me and my amplifiers.
I have recommended that ever since.
i.e. use factor of >=sqrt(2).
well, I suppose using any fan is no good in regard to components' layout and pcb design at all?! Does it make any distortion/ oscillation?
I will reconsider the case, perhaps monoblocks will solve the issue 🙂
sounds a bit suspicious, though I've got the point.
I would stay with myRevC design, plenty to be read in separate thread here. I'll try to find any available pcbs or alternatively get these ones. Seller confirmed that pcb comes with schematics. Furthermore, it could be updated in accordance to myRevC thread regarding components, I hope. This one also could be used, looks like some old version.
Meanwhile, I'll search for some other small cases with more decent ventilation and heatsinks 🙂
Cool! MyRefC monoblocs would be really deluxe.
With 4 ohm speakers you'd be shopping for. . . transformer spec comes out at an identical pair of secondaries in a range of 18 to 20 vac and at least 3 amperes. A 22+22vac transformer is too high in voltage--use about a step less voltage.
Okay, now at monobloc, you can possibly go without a fan if there is really significant intake ventilation at the bottom of the enclosure to allow cool air to come in and fairly good output ventilation at or near top of enclosure to allow hot air to exit.
That silent and free breeze is slower than a fan, so you do need to think about heatsink, as is true of most class AB amplifiers.
It is also necessary to keep capacitors cool for good longevity.
----
The prospect of decreasing non-audio workload is a lot of little steps, but it can add up very nicely. . .
An important step for cool running is super-quiet rectifiers. Mark Houston's photo is excellent: Synergy - LM3875 Gainclone Amplifier (Chip Amp) Rectifiers done like that help a lot!
And, 10uF at one or both rectifiers -or- a single 4.7uF 250v cap spanning across the rails, can remove a bit of heat inspiring noise. There's usually a slight difference in sonic signature, in the helpful, slightly less forward, direction.
At the chip, and in its datasheet, there's a 220pF cap from in+ to in- for RF blocker and you might check this as well as the MyRef documentation to see about using it to decrease non-audio workload, thus slightly cooler running. The little cap also usually assists stability and more stable is cooler running.
Speaker output zobel could be a conservative value like 47nF series to 5.6R; and, the not-usually optional inductor//resistor (see datasheet) can also help slightly.
Last edited:
I changed R6 to 10k and my offset is now reduced by around 4mV it's now around 8mV and it has also moved the cutoff of the input ahead of the feedback network as suggested by dtses. I can't make out any difference in the sound though.
You reduced the offset of a parallel amplifier? Well, that is good news.
Are all of the gain setting resistors precision match via digital ohmmeter?
The mute circuit is pumping a little bit of voltage and current into signal star (maybe causing your 8mv offset) and I would suggest that ground for mute cap be moved to 0v power ground instead of signal star. You do have a groundlift resistor for signal star and thus you don't want your mute cap pushing it lopsidedly.
If it is easier, you could defer the groundpoint for mute cap to the other side (power ground side) of the signal star groundlift resistor so that the current does not pollute the groundlift resistor. That cap just has the wrong spot for ground.
P.S.
+input cap versus -input cap: Since we're not trying to make a bass blocker, just make the -input cap bigger instead of reducing the +input cap smaller than ~2.2u. 🙂
P.P.S.
Some of the large multi-parallel projects like PA150, BPA300 illustrate the use of a trimmer so as to zero the offset. These push the signal star either farther or closer to center, depending on where you turn the knob. It looks a lot like the mute cap groundpoint error on your schematic, except that an offset trimmer is not stuck at just one lopsided setting.
Yes, I matched some 1% mfr with my multimeter for the RI and RF resistors.
I had taken the national's board artwork as a reference there also I find that the mute ground is connected to the signal star, but they are not using a timer cap . I can move the mute cap ground, it's just that have to do it with a wire, since the boards are already fabricated. The biggest cap I can use with the footprint I have on my board would be around 100uF / 63 and I believe these caps need to be really good quality stuff .
The 8mv is nothing to worry about if it is coming from both cores equally with identical voltage. You could measure that if you could measure the speaker output resistors individually.
thanks again, Daniel!
Noticed few posts that Revc doesn't produce so much heat (perhaps on 8 ohm speakers) while reading the building thread. Unfortunately didn't save the links to show posts, though anyway I'll bare in mind your advices regarding heatsinks very seriously! Being quite stubborn I would eventually try to stay with 2 channels pushed into some tiny (140mm x 60-80mm x 250-330mm) box 🙂
I would stay with >160VA toroid, 2x18 4.44A for both channels. So it's only 160VA what makes me feel that it should be 2x20k uf on the rails or 10k should be enough?
In case overheating will make me feel the chassis is boiling, I would simply convert it to monoblock, though this is not the way I would like to build it as the upcoming plans would be to build 2nd identical stereo amp and drive my ASW Genius via biamping.
you probably meant LM3875's attached to the chassis or which rectifiers, which pic?
myRevc has no zobel as far as I've seen. Perhaps it was in some version?!
That PCB is not a My_Ref... it's a simple buffered gainclone.
That one is a My_Ref clone and, apart ethical problems, works fine and is very well made (I've one).
If you're willing to build a My_Ref you can contact LinuxGuru (he has Rev C boards and kits) or wait for the GB of my variation (we're both authorized by Mauro Penasa, the designer).
10000uF are enough, the only version that uses 2x10000uF per rail is the My_Evo (not public and no longer available).
My_Ref don't use zobel networks in any of his versions (both official and unofficial), according to the designer they would impact performance and stability.
Last edited:
that's not a particularly small box for 2channels of power amp.
I used a 172x289x52 internal size, for 5input remote controlled integrated 60W+60W amplifier.
ALL of the audio was fitted into the rear 70mm or so (<0.7litres). The internal heatsink (160x50.8x40mm) took up the next 40mm. The power supply and digital control, etc took up the front 179mm that remained.
This was very much a case of assembling the modules, getting it tested and working.
THEN deciding how to package it and find a chassis to enclose it.
You can 10kuF, resistor (or cable), 10kuF, per each rail for CRC.
The first 10kuF section at the bridge rectifiers provides smoothing to create DC.
The second 10kuF section, at a point extremely close to the amplifiers, provides a tank section as well as a stable spot for speaker negative.
This CRC is compatible for either a separate power supply box or an compact/integrated build.
Have you considered Class D for bass amp, LM1875's for treble amp?
Sir, does that have enough ventilation and heatsink to support dtses's 4 ohm speakers with LM3886 amplifiers?
The heatsink allocation is nearly 56% of the total audio allocation.
I do not think I have compromised my own advice where I suggest doubling the National guideline heatsink dissipation.
I do not think I have compromised my own advice where I suggest doubling the National guideline heatsink dissipation.
thank you! I need to contemplate the layout inside the box, consequently deciding on dual mono config or My_Ref clone pcb.
Your version looks splendid, when do you think the final version would be available?
Initially a stereo power amp based on LM3886 according to myRev schematics will be build. Then if it's really so good, perhaps another one will be build 🙂
Btw, would it be ok if I use 2 equal toroids (160-200VA, 2 x 18v, for instance), but different brands (possibly the sizes would differ slightly too) for 2 stereo power amps?
You wouldn't want a slightly different tone per right and left, therefore, I think that you won't want differences in right versus left amplifiers.
- Status
- Not open for further replies.