[BoyElroy]

Alright, I may be totally out of my mind, but I came across something recently that I wanted to run by my esteemed fellow DIY'ers.

I rebuilt my Gilmore PSU and tried many different approaches to taming the heat/oscillation issue with the opa548. While adding a non-inductive 10 ohm resistor in series with the output caps works, I felt that when used in the psu of my Gilmore preamp (driving my Apogee stage speakers) it veiled the sound a bit and smeared a lot of detail.

The Tweak:

As a workaround, I removed the 10 ohm resistors and placed a single .120 uf poly cap between the -V in (pin #2) and ground. I did this only for the negative leg.

The result:

The bass was tighter and seemed to go lower. The highs were especially affected, resulting in a noticeably more transparent and clearer soundstage. In general, it sounded as if a layer of muddiness had been removed.

All the heat issues disappeared as well. The opa548's and 317/337's cooled down and things got back to normal. I'm using 5 uF poly caps before the opa548's and 1 uF Hovland Musicaps as my output caps.

Question:

By coupling the -Vin pin of the opa548 (negative leg only) with ground, would anyone know how this affects the current output? When using smaller caps (.01 uf caps from Radio Shack), the oscillation actually manifests itself as an audible high-pitched whine. Larger caps (.5 uf to 1 uf) , don't really stabilize the oscillation it seems as the negative leg opa548 overheats.

So, the .1 to .120 uf caps seem to do the trick, resulting in much better sound. I'm not sure, though, whether what I'm hearing has any grounding in measured performance or whether I'm seeing and hearing Pixies, as it were.

Thanks all!

 

[antness]

I think you're expecting a negative change in the sound by adding the 10 ohm resistor, which wouldn't seem to have a postive effect on the sound. I dont notice a difference in the sound either way. But to me, having a more stable, cooler power supply can only be a benefit to the sound and performance of the amplifier.

 

[BoyElroy]

Hey Antness--

Yeah, I agree. That's what's neat about this tweak; it keeps the amp totally cool and stable without the need for putting in a 10 ohm resistor.

 

[BoyElroy]

Well, here are some new measurements and impressions:

The following temps. were taken from my Gilmore PSU with KG's following recommendations on bypass caps;

1) .1uF from pin 4 to ground as close to the opamp as possible for
both opamps.
2) same thing with pin 5 to ground.
3) .1uF from pin 2 to 6 on negative supply opamp.
4) doing the same for the positive supply cannot hurt. (**please see comments below**)

negative leg LM337 = 52 degrees C
positive leg lm317 = 41 degrees C

negative leg opa548 = 32 degrees C
positive leg opa548 = 38 degrees C

The LM317/337 have 5 watt rated heatsinks and the opa548's actually have larger 8 watt heatsinks on them.

After two hours, the temps. across the board were slightly higher than with a single .1uF cap from the negative rail opa548 pin #2 to ground, but well within design limits for all the regulators.

**The one problem I had was adding a .1uF cap from the positive opa548 pin #2 to pin #6. This caused some oscillation and the temp. on the LM337 shot up past 78 degrees C. Removing this bypass cap brought everything under control again.**

In terms of sound, I am doing all my sound tests with the Gilmore set up as a preamp. My unit also has a stepped ladder attenuator, which I believe contributes significantly to the sound quality. That being said, when using my headphones (Senn HD-25-1's and Grado 325's) I cannot tell the difference between power supply cap configurations. When played through my McCormack DNA1 dlxe amp and Apogee Stage speakers, however, it is not that difficult to pick up on certain differences. My immediate impression is that KG's recommended configuration above sounds the best. The bass is clear and tight and the midrange full. The highs are extremely clear and smooth and open. I definitely recommend this setup over the 10 ohm resistor in series method.

Not to beat a dead horse here, but I think the Gilmore really benefits from a high quality stepped attenuator. The circuit is very quiet and of sufficiently high quality to benefit from the upgrade. Then again, I suppose you could say that about any pre/headamp...

 

[Doh]

Hey everyone,

Sorry to resurrect and old thread, but I just wanted to share my experience with the gilmore power supply and the troubles I've been having with oscillations since building it.

Ever since I built my gilmore amplifier, I've been having a nagging problem with the negative rail opamp. For some reason, it seemed to be getting hotter than the positive rail opamp and the voltage regulators even with power supply decoupling. It did not get extraordinarily hot...just warmer to the touch. This behavior occurs even with the two 5uF output capacitors removed. The amplifier seemed to be working fine. Yet, I found the behavior odd. Here are the steps I took.

1) Adding decoupling caps to the positive and negative voltage rails of the opamps helps somewhat, but still the asymmetry exists.

2) Adding a 0.1uF capacitor between pin 2 and pin 6 of the negative rail opamp causes the temperature of the opamp to skyrocket out of control.

3) On a lark, I removed the two ceramic capacitors in position c3 of the power supply. I figured heck, they're capacitors too, right? Anyhow, they're not in Gilmore's original schematic and from the layout, appear to be in relatively the same positions as the original 5uF capacitors. The result? Problem fixed! The temperature of the two opamps is now equalized. They're warm, but comfortable to the touch.

I haven't had time to thoroughly listen to the amp since I made the modifications, but I swear that there has been some recovery of the high frequency characterisitics and a great deal more clarity from it. Incidentally, after removing the C3 capacitors, I added in two 10uF 63V cerafine capacitors I had lying around. The amplifier still seems to be working quite well. No asymmetry in the temperatures of the opamps to the touch.

Doh

 

[Arzela]

Position C3? Where did these tie into?

Incidently, my power supply, with no
capacitors other than the four main
electrolytics, seems to work fine;
although the positive 548 and regulator
seem to run a bit warmer than thier
negative brethern.

 

[Sovkiller]

Quote:

Originally posted by antness I think you're expecting a negative change in the sound by adding the 10 ohm resistor, which wouldn't seem to have a postive effect on the sound. I dont notice a difference in the sound either way. But to me, having a more stable, cooler power supply can only be a benefit to the sound and performance of the amplifier.

Is there any chance that these mods were included in the upcomming new V2 gilmore?

 

[Doh]

Position C3 is where antness specifies two ceramic disc caps in the power supply. They appear right before the two final 5uF film caps. I don't believe they show up on Gilmore's original schematic.

My issue was the apparent temperature asymmetry even with the two final 5uF caps removed. It appears that in the V2 Gilmore, that these caps have been omitted, at least from what I can see from the pics. According to the headwize article, they're for the "lunatic fringe" They seem to be the root of many power supply oscillation problems if low-inductance capacitors are not used.

The cerafines are a bit of a kludge. I think their use is probably sub-optimal, but I had them lying around.

My setup is a little different from that found in Antness' kit. Instead of the generic 4700uF 35V caps he specifies, I use 3900uF 35V Panasonic FC series electrolytics. For the 5uF capacitors, I substituted 5.6uF panasonic film caps instead of the WIMA box caps in the kit. They just fit onto the board with a little lead bending.

Incidentally, I also noted a bit of a DC voltage (around 12-15mV) at the output of the amp before the mods. This was actually more concerning to me than the temperature issue. After the mods that the final DC voltage offset for the output of the amp dropped from a measured 15mV to nothing. A nice side effect. I believe that the professor specifies that anything under 25mV is acceptable, but I can't help but think that less is better here. (The voltage should be zero with a working servo, no?)

 

[Jupiter]

Does the temperature problem occur when the power supply isn't powering the amp? I just finished the power supply and without the amp connected the temperature on the opa548 seems fine. (I haven't finished the amp yet.)

I've used most of the parts antness suggested including C3.

 

[mhamel]

I just finished up a Gilmore PS based on Antness' board, and I had issues with oscillation as well. I used all of the suggested components from his parts list, but as soon as I turned the power on, the temp of the negative opamp would shoot straight up, and the positive opamp would rise a bit more slowly but still way out of any comfort range.

I added the caps per KG's suggestion earlier in the thread and everything's running well now. No more oscillation, both regulators and opamps are slightly warm, but nothing out of line.

-Mike

Edit: Ok... nevermind what I just said. I just took the PS out of the enclosure so that I could add feet to the bottom of the it, put it back together and now the negative opamp seems to be oscillating again. Argh.

 

[Doh]

When my power supply first started oscillating, some really nastby buzzing started coming out of the amp. The LED's looked like they were flashing at a very high rate. Even when the amplifier seemed to be working, there was a lot of asymmetry in the temperature of the regulators and the opamps. At times, the negative opamp would be warm, while the positive opamp was virtually cold. I think the state we're striving for is a little bit of warmth with both the opamps and the voltage regulatorsl, which is what I get now. Here's a brief synopsis of the mods I went through to get things working seemingly right...

1) Full gilmore power supply. Oscillated like mad. Buzzing from the amp. Power supply seemed all-right without the amplifier board connected.

2) Decoupled the positive and negative rail opamps. No real help.

3) Added the pin 2 to 6 0.1uF capacitor to the negative rail opamp. Temperature of that opamp shoots totally out of control. Still buzzing from the amp.

3) Removed the negative rail 5uF output capacitor from the power supply, leaving the positive rail capacitor in place. Amplifier has SOUND!!! Stayed with this configuration for a few weeks. Was bothered by the asymmetry of temperatures between the negative rail and the positive rail opamp. Regulators and positive rail not warm enough so that touching them for a prolonged period of time was unbearable.

4) Decided that the temperature asymmetr can't be normal. Likely the negative rail opamp is oscillating. Measured the DC voltage from the amp. 15mv or so. Removed the other 5uF capacitor. Temperature stabilizes. Everything running just a little warm, but not too hot to touch for prolonged periods of time.

5) Tried adding in the 0.1uF capacitor between pins 2 and 6 of the negative rail opamp. Temperature skyrockets out of control, but amplifier has sound, at least for short periods of time. Leaving it on causes loss of sound again.

6) Removed the capacitor between pins 2 and 6. Everything seems ok. Measured DC voltage. None.

7) Added electrolytics caps as the output caps. 10uF each. Everything still seems ok. Nearly symmetircal temperatures in regulators and opamp. Still no DC at the output. Bottom line, I like it. This configuration will stay until I decide to put in some other caps at the output of the PS.


I'm sorry that I can't offer any advice here, but I hope that a little more information may help people in building their amps, or help us get to the bottom of what is going on.

 

[mhamel]

Well, I decided to try electrolytics at the output, and so far, so good. The smallest ones I had on hand were 33uF Nichicon bi-polar caps. I added a pair of 3.3uF Panasonic film caps and everything seems to be running nice and cool and stable again.

-Mike

 

[Doh]

What kind of panasonic film cap are you using? Do you have a part number?

 

[mhamel]

They're ECQ-E Series.
3.3uF, 250v

Digi-key part number P10985-ND.

Datasheet:
http://www.maco.panasonic.co.jp/www-...BD0000CE34.pdf

I just happened to have a few in my parts bin... don't remember what I even bought them for originally.

-Mike