[JamesL]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif - The new class A biasing scheme removes previous requirement for hand-matched transistors - The new biasing scheme is adjustable, with a single pot adjusting both channels' bias - 4-layer board, allowing full ground plane and heat-sinking for the buffers - better bypassing

Thanks for the additional notes =)
I mentioned the changes in biasing scheme but I wasn't completely sure what it did.
Does the bypassing refer to the addition of c3/c5? The first time I looked at them, I thought those were the electrolytes, but I realize they were moved prior to the rail splitter.

And lastly, do you think the ad8610/lmh6321 will perform well on a 9.6v supply? The datasheets specify min Vs of +/- 5v for both, but it seems theres quite a bit of leeway before the opamp completely craps out.
If it's a bad idea, any alternatives that you can recommend in particular?
Thanks

 

[diditmyself]

LMH6321 is specified from 5V so the use of a single 6F22 battery will cause no problems at all. Since the output transistors are common collector you'll loose about 2V swing but since it's the same for most of the good sounding opamps it doesn't matter. It might bother some half-deaf loudheads with high impedance phones.

Regarding the cap discussion: don't worry about the cap size, you wont hear any difference anyway. If you're not convinced do a blind listening test with normal and "oversized" caps in any 3-channel amp.

 

[diditmyself]

... and i forgot the AD8620. As far as I remember it's one of the better performers at low voltage supply, but I don't really like the sound of it and it's far too expensive to recommend.

 

[hopeless]

Quote:

Originally Posted by diditmyself /img/forum/go_quote.gif ... and i forgot the AD8620. As far as I remember it's one of the better performers at low voltage supply, but I don't really like the sound of it and it's far too expensive to recommend.

The LT1469 and LT1468 sound better than the AD8620 and AD8610.

Also it can work great with the LMH6321 in terms of tonal transparency. Try to believe.

 

[joneeboi]

After much delay, I have cased up my PIMETAv2. I haven't implemented Class A biasing yet (where are my transistors?), but it's good so far. DC offset used to run wild (think ~100mV), but after throwing in my bass boost, it seems to have calmed down. More info will come after work.

 

[Voodoochile]

That's great man. I still haven't cased mine, but I did make the PSU for it, and I have a small case on the way.

Looking forward to seeing your cased up.

 

[joneeboi]

Ehh, it's nothing to write home about. I'm actually just lucky that everything fits inside. As for a power supply, I just have the AC-DC wallwart putting 24V through an LM7818.

Just a note about bass boost. In the schematic, there's a switch to turn BB on and off, but there really isn't a need for it if you just turn the bass boost pot all the way down. Saves money, panel space, so on and so forth. I don't know if this has been covered in other BB discussions, but this is the first time I'm using it.

Also, does anyone have any explanation for why my DC offset was running wild like that? For just one channel, it would go from 38mV to 120mV and just bounce back and forth when the music was on. I'm just connecting my iPhone as source. It's fine now that my BB is wired up. I'm BrownDogging AD8620 and AD8610 into a mostly SMD build.

When I get the chance, I'll go into more depth about my LM7818 config.

 

[tangent]

Measuring DC offset while music is playing means nothing. Your meter is just being confused by the AC.

 

[joneeboi]

Fair enough. Now what if it varies with the pot and no music is playing? A fully open volume pot meant no DC offset, but as I decreased the volume, the DC rose.

 

[tangent]

It could indicate a problem. Are you using the new values on the schematic, particularly R7 = 100R instead of 1K?

 

[joneeboi]

My R7 is indeed 100R. I used 3.32K for R1G and R4G, but I figured it would be okay since it's just a voltage follower anyway.

 

[tangent]

You should have the inputs grounded when doing this test. If the inputs were floating or connected to a source, your results are again questionable.

 

[joneeboi]

Hm, I can't remember if I did ground the inputs for those tests. I think I did, but it doesn't matter now that my DC is pretty much zero.

Anyways, on to the main event. I finished building my PIMETAv2 last night in preparation for the Calgary meet this Saturday. I installed Class A just yesterday night, and biased my RBLIM of 287R to have ~0.56V across it, about 2mA of bias. I tweaked my BB to only have the pot and caps in parallel (no R4, so it's in unity boost), so it's plenty deep right now through my normally airy SR60s. I must say, I may be a bass junkie.

My pot is 50K, and I have 0.22uF K42Ys from my Millett Max build paralleled with an additional 0.1uF of Wimas I grabbed from Mouser. Wow, I can get used to this much bass. All these values weren't really calculated. Class A could be biased deeper, but I don't know if I really need to; I could stand to with the additional PS headroom. The bass sounds great to me, so I have little impetus to get it "just right."

Here is my configuration:

- SMD build. All resistors are SMD, and so are C3, C4 and C5. It was good experience, but I much prefer having through hole. It's easier to see what you installed because it's all on top of the board.
- AD8620 and AD8610. I used to have OPA627s with my PIMETAv1.1, but i think I might have blown them up
- R7s are 101R
- R4 is absent. I have the amp set up in unity boost.
- C4 is 1.5uF
- Class A 2mA. Trimpot 5K and RBLIM 287R. Both my LEDs have Vf 3.5V.
- I went with the 1.1mA CRD for RLED
- Volume pot is Alps Blue
- BB pot is 50K Alps RK097
- Jacks are nothing special
- No switches for PWR and BB. If I'm not using the amp, I don't really care to shut it off. If I don't want BB, I just turn down the pot. It turns on when I plug it in.
- Case is 1455K1201.
- PSU is a 24V AC/DC wallwart into an LM7818. Here's the different part for me. No particular reason I did it besides proof of concept. You install it the same way you'd solder in the LM317, but you don't connect any diodes. Instead, you bridge their anodes (non=striped ends), and connect pin 2 with V-. You can use the other end of RCCS to do this if you don't want to solder directly to the LM78xx. If you just compare the two pinouts with the schematic, you'll see what I'm talking about. I've linked them for your convenience. Plus, you have to make sure you include the proper stability capacitors. I don't know if tangent plans on incorporating this into the next design, but it can be done with this revision of board if you so desire. I also got a huge heatsink for the VREG, which I don't think helped with the casework. Nevertheless, huge heatsinks are always cool.

If any of you are familiar with my workmanship, it's not as it should be. I usually only use one type of hookup wire and I don't have any heatshrink. Just lazy to invest in the proper tools, but I'm getting there. I'm a little hesitant to show the internals, but since I'm a prototyper and not just doing my own thing, I'll share. You'll understand when you see it.

Anyway, just wanted to report a successful build with no major complications. If anyone wants my BOM, I can share that too. Thanks, tangent, for including me in the proto group. I really liked the card. Classy.

#0006, done.

 

[linuxworks]

I had to fix my pm2. I had a power supply failure on my 20-24v rail. oh my.

I think the 2 op-amps survived but the 3 buffers were on their last legs

ground channel was 'gurgly', like under water sounding. bypassing ground channel (taking ground from input ground) was mostly ok.

8620 had problems at various voltages, yet 2134 was ok almost all voltages.

at lower listening levels, things were passable but as you turned the volume up, things distorted a lot.

I replaced all 3 soic buffers. that fixed it.

while I was in there, I put in the tangent fixes from v2.00 to 2.01. I think I got most of them:

r7 goes from old of 10k to new of 100
r4g goes from old of 4.7k to new of 4.32k
c6g goes from old of 10pf to new of 4.7pf (I have not done this one yet)

for now, I've left out r11 (l.r,g) entirely.

I had extra buffers so I just clipped the old ones off the board (ouch!) and then soldered new ones in. first did l/r and verified that fixed the main channels. then did the ground buffer and that fixed it all.

lesson: don't get power supply problems and you won't have to fix circuits downstream of them!

 

[joneeboi]

Where was R4G mentioned? Why was it changed?