When looking at the powersupply of the gainclone, I see only a very basic supply with a 1000uF cap, an oversized transformer here and there and that's all.
In tube equipment we do a lot to achieve the best powersupply, especially for SE-amplifiers. Soft-rectifiers, choke-input to decrease current spikes on the diodes, CLCLC-filter to have low Z, but as well nice ripple rejection while having a fast supply using small capacitors etc. Exotic components like chokes with amorphous cores, blackgates etc.
When looking at the JLH or Naim-clones, we discussed regulated supplies, capacitor multipliers etc.
Why is this not longer relevant when building a gainclone ? Obviously it would increase complexity a bit, but I guess the aim here is not to make compromises.
In tube equipment we do a lot to achieve the best powersupply, especially for SE-amplifiers. Soft-rectifiers, choke-input to decrease current spikes on the diodes, CLCLC-filter to have low Z, but as well nice ripple rejection while having a fast supply using small capacitors etc. Exotic components like chokes with amorphous cores, blackgates etc.
When looking at the JLH or Naim-clones, we discussed regulated supplies, capacitor multipliers etc.
Why is this not longer relevant when building a gainclone ? Obviously it would increase complexity a bit, but I guess the aim here is not to make compromises.
Blitz said:
I have used very basic unregulated power supplies (normal silicon bridge plus 3300uf caps) for a very versions of jlh1969. No problem whatsoever.
i think the "minimal list" approach to gainclone ps is ill advised. there are various simulations that indicated problems with small filter caps, and the psrr of the gainclone goes down pretty fast with frequency, especially on the negative rail.
for mine, I used the same power supply on my gainclone.
I doubt regulated supplies will bring us closer to nirvana. It's difficult to like the sound of regulators in preamps let alone power amps. How many tube power amps have regulators?
Amorhous chokes for a GC? Now that sounds like an idea
It should be able to pass several amps, right? Certainly not off the shelf.
Batteries otoh are a realistic proposition with the GC and some members report excellent results.
Amorhous chokes for a GC? Now that sounds like an idea
It should be able to pass several amps, right? Certainly not off the shelf.Batteries otoh are a realistic proposition with the GC and some members report excellent results.
Isn't the amount of capacitance on the PS relevant entirely to the power one expects to get from the GC. If your GC is never asked to give more than 15W out 1000uF/rail is just fine. If you'll be pushing it to 50W, by all means put 4700uF per rail (1000uF at the chip and 3300 at the PS section). I find it very flexible to that kind of tweaking.
Also (I don't have much knowledge myself, so it's more a question than answer) isn't the PSRR of the SE-amplifiers way smaller than what opamp (LMx875) would offer? I think that's another reason to minimize the PS by taking advantage of the good PSRR of the GC.
There is always an optimum or balance of how much you gain/lose and how much more complicated the circuit becomes.
/Greg
Also (I don't have much knowledge myself, so it's more a question than answer) isn't the PSRR of the SE-amplifiers way smaller than what opamp (LMx875) would offer? I think that's another reason to minimize the PS by taking advantage of the good PSRR of the GC.
There is always an optimum or balance of how much you gain/lose and how much more complicated the circuit becomes.
/Greg
Power supplies
As Joe Rasmussen pointed out, the various gainstages internal to the LM3875 appear to use active current sources (with the exception of the output stage) . This factor "suggests" that this chip will be less sensitive to the power supply effects.
However, contrary to what some members of this forum believe or claim to have experienced, I personally believe that a building a stable and quiet supply as possible within our budget(s) will go quite a ways towards maximizing the sonic qualities that this chip is claimed to be capable of. What I mean by this is I believe that noise filters on the primary side of the power supply transformer will help keep RF crud from entering the amp. Use decent size power supply capacitors following the the rectifiers. SOme claim that using what are considered small values will sound better. I have found the opposite to be true, at least in the cases where the load (your loudspeaker) has a pretty low impedance, notably in the bass range. Larger value capacitors (of good quality) will help maintain a consistant DC voltage. Too small a value will result in a sagging of the power supply rails momentarily (in the presence of LF signals). There is some theory that using a large cap as opposed to a small value cap will have the effect of not having the supply respond fast enough to certain signals. A suggested solution is to use multiple small value caps which total the desired larger value.
My own experiments have proven (to my ears) that small value caps are inadequate with the dynamic speakers I typically use (restored Allison 1's, old Infinity Qe's, some Innersound ESL's and soon a variation of the Linkwitz Phoenix-Orion) 10,000uF I have found to be the smallest value that I don't recognize any sonic effects. Driving the Innersound ES panels with paralleled LM3875's I find that I can use the 1000uF that many are using with no noticable change from 10,000uF. But, the ES panels have a high impedance at lower frequencies and they are only active at frequencies above 200 hz (400 is the original crossover point)
I guess my point to all of the above is to start your amp project using the recommended capacitance with your intended speakers, then add capacitance and see if there is a change.
If anyone is curious about using high current regulators on their Gainclone, they might want to pick up an old AUdio Amateur from around 1981. They modified a Hafler 200 and added regulators to the power supply rails. The article included information on adjusting the regulator output for use with different amps. If there is interest I'll see if I can find it and post a schematic.
-Tom
As Joe Rasmussen pointed out, the various gainstages internal to the LM3875 appear to use active current sources (with the exception of the output stage) . This factor "suggests" that this chip will be less sensitive to the power supply effects.
However, contrary to what some members of this forum believe or claim to have experienced, I personally believe that a building a stable and quiet supply as possible within our budget(s) will go quite a ways towards maximizing the sonic qualities that this chip is claimed to be capable of. What I mean by this is I believe that noise filters on the primary side of the power supply transformer will help keep RF crud from entering the amp. Use decent size power supply capacitors following the the rectifiers. SOme claim that using what are considered small values will sound better. I have found the opposite to be true, at least in the cases where the load (your loudspeaker) has a pretty low impedance, notably in the bass range. Larger value capacitors (of good quality) will help maintain a consistant DC voltage. Too small a value will result in a sagging of the power supply rails momentarily (in the presence of LF signals). There is some theory that using a large cap as opposed to a small value cap will have the effect of not having the supply respond fast enough to certain signals. A suggested solution is to use multiple small value caps which total the desired larger value.
My own experiments have proven (to my ears) that small value caps are inadequate with the dynamic speakers I typically use (restored Allison 1's, old Infinity Qe's, some Innersound ESL's and soon a variation of the Linkwitz Phoenix-Orion) 10,000uF I have found to be the smallest value that I don't recognize any sonic effects. Driving the Innersound ES panels with paralleled LM3875's I find that I can use the 1000uF that many are using with no noticable change from 10,000uF. But, the ES panels have a high impedance at lower frequencies and they are only active at frequencies above 200 hz (400 is the original crossover point)
I guess my point to all of the above is to start your amp project using the recommended capacitance with your intended speakers, then add capacitance and see if there is a change.
If anyone is curious about using high current regulators on their Gainclone, they might want to pick up an old AUdio Amateur from around 1981. They modified a Hafler 200 and added regulators to the power supply rails. The article included information on adjusting the regulator output for use with different amps. If there is interest I'll see if I can find it and post a schematic.
-Tom
Amplifier regulators
Here you are. I'll have to dig up the component values. (post those later) But, the schematics give some idea of the circuit complexity. In the article they note the results (max power output, distortion etc) on a Reg. Williamson power amplifier. The unregulated version put out 44 watts (8ohm) and 52 watts (4ohm). With this regulator it put out 62 watts (8ohm) and 81 watts (4ohm) . Distortion decreased by better than 50% over the unregulated version.
Now I know that many feel that specs mean nothing, but, the numbers at least give an idea of operating performance.
Oh yes, the power devices of the regulator must be properly heatsinked.
tbla said:
yes - please.......
Here you are. I'll have to dig up the component values. (post those later) But, the schematics give some idea of the circuit complexity. In the article they note the results (max power output, distortion etc) on a Reg. Williamson power amplifier. The unregulated version put out 44 watts (8ohm) and 52 watts (4ohm). With this regulator it put out 62 watts (8ohm) and 81 watts (4ohm) . Distortion decreased by better than 50% over the unregulated version.
Now I know that many feel that specs mean nothing, but, the numbers at least give an idea of operating performance.
Oh yes, the power devices of the regulator must be properly heatsinked.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
heatsinked
In tube amps it's common to replace capacitance with inductance and it really works well. I am not sure if anyone has seriously tried this with SS - there seem to be some CLC power supplies in the Pass forum but i've never read a subjective opinion regarding their use - it seems the designers are preoccupied with ripple rejection. Any experiences?
Yes, there are, look at the work of MArk Levinson with his Cello LAbel and Erno Borbely who clearly states that only choke-input filters really bring you to nirvana. Download from www.duncanamps.com his powersupply simulator and look at the current spikes which will be produced by a normal pi-filter and you know why these two gentlemen like to use chokes...besides all the goodies in filtering out RF-garbage, energy storage inthe chokes etc. Nevetheless they are big and expensive, but so what.
Well chokes could make sense as passive parts before the regulation to avoid noise and get low ripple with loe DC-resistance. I think the NAim-clines as well work more or less in class B and there the discussion clearly was how to get the same nice sound from the regulators Naim use in DIY and not to use only one 1000uF cap.
I know one thing for sure though. Whenever I used those AC filters on primary (in amps), the sound degraded substantially.
I even stopped using them in DACs. Initially, I thought they were improving things substantially, but I'm not sure about that anymore. The sound without AC filtering is somehow more convincing (at least to me). But it probably depends largerly on mains quality as well. I think in my area it's not that bad.
I even stopped using them in DACs. Initially, I thought they were improving things substantially, but I'm not sure about that anymore. The sound without AC filtering is somehow more convincing (at least to me). But it probably depends largerly on mains quality as well. I think in my area it's not that bad.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.