Need suggestion for the improvement of Petter gainclone 3875 kit

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
My current set up is

Plitron transformer: 1000 VA dual secondary 25v , primary 110v ( 35.6v after rectifier)
premium kit. (Actually basic kit but i change all to Caddock registors)

I tried Calo snubber and like it work nice when you have a smaller transformer but i can feel that it lose something in the music and like the original after listening to many many hours.

I wanna try change the pana cap to BG 100 uf N type close by the chip pin and put the pana 1500 uF and 4.7 uf Blackgate N type in front of the rectifiers.

My question is: Is it a good upgrade or any other suggestions

Thanks all.

The question is

The question is what sort of alteration would you like to make in the amplifier.

What do you want it to do differently than it does now?

Do you want a smoother sound? If so, big carbon for resistors that pass audio information and/or possible addition of a "warm sound" input filter cap. For its location, see the LM1875 PDF at National Semiconductor.

Do you want less heat? If so, the small resistors are fine, and just make the input load add up to 10k (as in an additional 20k via an A20k pot or a plain resistor). And, use less voltage whenever this layout is used, else the amp will peak at the high midrange.

Do you want a wider bandwidth? If so, speed up the chip by increasing the NFB's 22k up to 56k (the 1k doesn't change--unless its mistakenly 680 ohms). Also increasing the input load resistor to 56k (input in-series changes to 2.2k or more and it should be carbon). Increasing the speed fools the limiter, reduces the clipping, increases the heat, and increases the power output, while potentially decreasing the DC output (if you have the NFB cap installed).

Do you want a wider audio presentation? If so, use at least 35+35 volts DC, and correspondingly braced-up thermal management. I much prefer this approach over the low power approach.

Do you want a smoother treble? If so, block solid state hetrodyne (block ultrasonic weapon noises) by installing 0.1uf economy polyester (not polypro) capacitors across the speaker terminals.

Do you want a professionally regulated amplifier?
Either use a preamp with tailored corrections for the power amp stage or skip the preamp and. . .
If so, tune it to taste by component choice. Paying greatly for resisters will get you resistors that make the amp run cool, possibly put out a lot of dc offset and make a screech in the audio. This is often mistaken for added detail, but it is an unlevel frequency response. DC into the speakers can sound warm, but its just masking instead. Very bright amplifiers can use "warm" resistors and capacitors to level the response. Also regulate the bandwidth to personal needs.

The issue is more severe for LM3886 and almost non-existant for LM1875. You've got a nice middle ground there with LM3875.

See my post here about a recent adventure with a similar but much brighter-sounding LM3886 kit.
Thanks Daniel for your suggestion . I know the title is a little bit misleading though. Actually, i like the set up now but as i remembered some people tried to replace the cap with 100 uf and push back the pana cap in front of the rectifier but i wanna see the public opinion on this before i make a change. Anyway, thanks a gain for the reply

There's probably no need to worry about the kit's power supply layout. The amplifier chips have 95db or so ripple rejection.
So, a change in an already good power supply is expected to be inaudiable.

There are places to make audiable changes. Thus, my response about those instead. ;)

The LM3875 is a hi-fi amplifier wheras LM3886 is a television amplifier, so do bear that in mind when you review the mod for LM3886, which may or may not be helpful, although restricting bandwidth to audio, does give more power.

If you were talking about the NFB capacitor, then 47uf is fully passing the audio bandwidth already.
A larger value will waste power (irritate Spike) by attempting to increase naturally present Theta waves.
However, a physically larger cap at the same value will usually have lower ESR, which is certainly audiable.

The means to apply these is "in proportion" with the circuit diagram/examples compared side-by-side.
The mod referenced is where I had grafted the support components for LM1875 (HiFi), a reference document from another company, lobotomized the Spike, restricted amplification to audio-only, added support for plain ordinary speaker cables, and supported using a volume control potentiometer instead of a preamp.
And then a preamp was no longer necessary because the chipamp is under control.

In comparison, a power supply mod isn't too audiable. ;)
Oh, more science? Okay: ESR and Capacitor Sound

Most of the audio effects from caps either come from signal stability or ESR (a form of added resistance).

Before purchasing extraordinarily advertised capacitors, there's a reasonably inexpensive thing ($60 to $300) called an ESR meter, which will tell the efficiency of a capacitor.

Most television technicians own an ESR meter.
Some perfectly ordinary capacitors document their ESR and their ripple/speed/signal stability genre.

When, ESR, capacitance, and speed match between any two capacitors, they will sound identical and then it does not matter if one of them is filled with snake oil and plated in gold.

So, if you can get the figures for the two capacitors that you're considering, then you can know for sure if there will be a change. You can compare the figures for capacitors reviewed to sound warmer or colder to find out what sort of change you can expect.

Compare the empirics to the specs and you'll get priceless tools on how to make purposeful modifications.

For instance, a cheap and tiny polyester cap has a very high ESR. This is similar to a great quality cap IF it also had a resistor on both legs.
Thus, the cheap and tiny 0.1uf polyester cap may be placed across the speaker terminals inside the amplifier enclosure, where it will help folks that have mistakenly left off their zobels.
The reason for leaving off the zobels is because the manufacturer specified polyester and when it is tried with polypropylene (less resistance), the effects are too severe (having subtracted 8 ohms of resistance from specs). Whoops! ;)

If we were talking the NFB cap, a high esr model will decrease the gain the same as installing an extra resistor, and a high ESR model is less effective at concentrating the power into the audio band--so less bass.
A low ESR model is more efficient, and its effect is more "sudden" so that when this is precisely aimed at the bottom octaves, there is additional bass.
When the capacitor is less capable of speed, yet also low ESR, and it is passing an audio signal, this is a warm capacitor, just like a bass booster.
When the capacitor is precisely aimed at the bottom octaves, yet capable of very high speed and has low ESR, this is like a loudness contour.
Not so enjoyable in gainclone NFB: When the capacitor is high ESR (cannot be precisely "aimed" at a particular frequency), but is capable of high speed, this is a bright-sound capacitor--a treble booster. This also happens if a good high-speed cap is aimed at a too-low frequency.

The LM amps are hyper sensitive the the smaller of the two resistors in the NFB.

The effect of resistors onto capacitors can alter the cap's speed and certainly does alter their ESR. Thus, putting a high quality cap with a too-small resistor is out of spec for a chip that specifies ordinary capacitors. The accomodation to use the high quality (low ESR) cap is a larger resistor to "break even" and maintain correct totals on spec. I'm speaking of the usual 680 ohm mistake in the NFB of kits. The manufacturer spec'd 1k.

If you were to use a lower ESR (higher quality) cap than expected, then the necessary NFB resistor value would increase, not decrease.
For extra fun, just combine an extremely efficient, very high speed cap with a too-small metal resistor with a similar extra-bright nature. . . and a LM3886 becomes a military grade ultrasonic weapon, instead of an audiophile product.
This is the large nightmare of noise that I received in a small and innocent-looking boxfull of very elegant and durable parts.

Adding elegance had thrown the specs.
It was the beauty, quality, and elegance of Brian's kit (surprisingly similar to your LM3875 kit), that had me believing I had made a grevious error in assembly--TWICE!

Twice? I've a "bon ami" rep with soldering, as in "haven't scratched yet." So, two parts assembled wrong was hard to believe. I had also used the high-zoot solder and the special flux that if it gets on the bottom of your shoe, the shoe rips right off on the first step. Not a loose connection then. ;)

So it had to be something out of spec. I grabbed the ESR meter (that I made--that fired up on-spec at the the very first use of its power switch) and looked for clues.

The ESR meter and a study of the PDF's put the lie to the mis-assembly notion fast. The more-elegant parts, in some cases, had thrown the specs by threefold. The one compensation, 680 ohms instead of 1k, was completely backwards, according to National specs and the ESR meter (measuring both capacitor and resistor simultaneously).

With an ESR meter and some imagination, I was able to make corrections and employ the high quality 47uf NFB cap by changing the resistor to carbon (lower the speed of the cap to warm up the audio sound) and 1.1k (increase the ESR to correspond to manufacturer's specified support components).
This had decreased the gain, so I changed the NFB 22k to 56k to put the gain back up without altering the "sound" of the cap.
That irritated the amplifier because its input load and NFB should match (when a volume pot is in use). So, I changed that (input loader) 22k to 56k as well.
Now there was too much gain and a curiously cramped bandwidth, so I increased the in-series resistance in the input to restore factory recommended proportions.
Oh wow.
These higher figures had partially lobotomized the Spike so that he quit yelling at me. I could get double the output without interference from the unwelcome power limiter circuit.

The clues were the support components of LM1875 and my new ESR meter to decode the audiable effects of capacitor's inbuilt resistance.

This nature of resistors can be measured when hooked in-series to a capacitor and measuring its speed. A metal resistor will preserve the high-speed and maintain the inherent brightness of a given cap's sound.
The cap I was given was already too bright, which was put with a metal resistor that helps it stay bright or makes it even worse. Thus there is the subject of carbon, which will not maintain the cap's speed, turned all of the capacitors warm. In addition, carbon, itself is a screech remover because it is warm sound not possessing the qualities to operate well in the ultrasonic range. ;)

The math doesn't subtract all of the empirics because dumb luck had played a major role in my repair job--see the comments about the interreaction with the, now-sedated, Spike system.

The topic of capacitor sound involves comparing empirics to measurements for accurate predictions of results.

One of these measurements is ESR--a vital consideration wherever there is a specification for a resistor and capacitor in series to alter or pass a signal at a certain bandwidth.
Too "good" esr can throw the specs just as easily as a bad capacitor--either is an alteration of in-series resistance value.
Imagine substituting a ceramic cap where a polyester cap should go. What happens? Ten times off the specs. ;)

So when making improvements on audiable signal qualities, try to adjust in proportion so that bandwidth isn't altered.
I learn a lot from you too!

Oh, what a beautiful compliment! I learn a lot from you too!

What a wonderful clue to give--a warm and high quality resistor. Oh thank you!

I finally got my Technics SA-5370 restored with the gainclone kit. I've got to admit that after all the tube-tone and extra power tweaking on the test bench, that her massive EI core transformer made for one heck of a surprise when I punched the bass booster button. I've just never seen the grilles blow off the speakers before. It wasn't quite like the old memorex ad, but really similar!

At first, days ago, I didn't even think the kits would work more than 20 watts each, but after a disappointing false start, days of test bench, and a re-assembly. . . Oh my gosh!
The test bench has only a 1 amp 36va center tap that will agree to 4 amps when needed. The Technics SA-5370 has a cantelope size transformer resulting 40+40 DC.

So, after tuning the amps for level reponse and as much clean power as possible (on the small transfo), today's reassembly (on the big transfo+very modded PS) results in super clean tube-amp like sound that's strong enough to redline my PA power meter.

Curiously, at such high voltage, the carbon R's that smoothed out the tone also seem to have acted as a dynamics booster.

Classical is unusually authentic--sort of like point blank range. Rock and dance is quite similar to a discotech with a wooden floor. Her passive pre has even more carbon. The raspy is now all completely gone. The remaining typical gainclone "clarity effect" survived despite the level frequency response.

WOW. This can toast the reference amp for fidelity and low listening fatigue.

See picture. I will be adding more heatsink parts and also add thermal paste between heatsink and metal frame--also going to thermal paste the transfo.

P.S. I'm off to seriously tamper some LM1875's as soon as I can find some copper heat spreaders. So there's a limiter? We don't think so! ;)

P.P.S. OH MY GOSH!! I just now had a weird experience!!! My mother just came over to my house in the middle of typing this.
She pops in a european dance CD and cranks the sucker ALL THE WAY UP. Off go the grills again! Next, she's doing some kind of sockhop vs. disco moves all over the living room!!! I've just never seen her act this way before. Now she wants one too. ;) I guess that's a good sign.


  • newagainchipampmod 480.jpg
    newagainchipampmod 480.jpg
    57.5 KB · Views: 540
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.