[cobaltmute]

I should add that I'd be willing to take a stab at the layout. I've got some time over the next few weeks in the evening to give it a shot but I won't promise anything. It defintely won't be a one nite wonder

 

[TzeYang]

Quote:

Originally Posted by cobaltmute /img/forum/go_quote.gif If you remove the JFETs entirely, you've just made a Pimeta.

I'd be very happy with a uber small sized PIMETA.

 

[cobaltmute]

Listening to my Pimeta V1 and having just received my Pimeta V2 board, I'm not complain about the sound of a Pimeta

 

[TzeYang]

Maybe if you're going to remove the JFETs you can add in other luxuries such as on board jacks, and bandwidth compensation caps? The latter is really really useful.

 

[NelsonVandal]

With a hand on your heart, can anyone say for sure that isolating the input opamp with CCS's in an amp such as this improve the sound? And what about Class A biasing the input opamp, Jung-style multiloop or large capacitors in a 3-channel amp?

What I would like to see is base-stopper resistors between opamps and buffers, pads for compensating caps (from opamp output to inverting input), pads for a zobel network, on-board jacks (MX-387GL from Marushin is small, cheap and durable) and an on-board trickle charger and DC-jack.

One of the drawbacks in PPAS is noise. The high gain and largish input and feedback resistors make this amp noisy, and it's audible when used with IEM's. The high gain is also problematic with IEM's because the channel balance at the beginning of volume pots are bad. A gain switch would be handy.

The dilemma of designing a portable amp is that you need it to be able to handle low impedance phones, and the use of buffers make the amp large and power hungry. One option is to find a good sounding, high current opamp that operates at low voltages, and AD8397 has been dominating in this category. The problem is that it doesn't sound that good and runs at a high quiescent current. Maybe we should put our heads together and start from scratch and make a "Head-Fi portable high performance amp".

 

[tcpoint]

I haven't noticed a high level of noise with the amp. It's been quiet - even when I turn it all the way up. However, I set it for a gain of 5. I haven't listened to it with IEMs. I will do that this afternoon.

 

[NelsonVandal]

I have an original PPAS with LME49720/LMH6321, and it's noisy. Gain is set to 4. It's probably noisier than the new one because of the layout without decoupling caps in proximity to the opamps.

If I simulate PPAS with AD8620 (one of the least noisy JFET input opamps there is), I get a noise figure of 165 nV/hz. If I set the gain to 5, I get 75 nV/Hz. Without the multiloop and with lower resistor values (input and feedback) the noise is below 50 nV/Hz. With a low noise bibolar input amp like LME49720 it's possible to get the noise a fair bit below 20 nV/Hz if low resistor values are used. Remember this is sim, and the noise is probably higher in real life.

To my ears the noise is quite audible with largish resistor values and high gain.

 

[cobaltmute]

That datasheet for the LMH6321 seems to imply (pg 9) that it gets noisy at low frequencies.

I just have to create a layout for the pot and then I can start working on laying out the whole board.

 

[NelsonVandal]

Quote:

Originally Posted by cobaltmute /img/forum/go_quote.gif That datasheet for the LMH6321 seems to imply (pg 9) that it gets noisy at low frequencies.

That goes for all amps. The noise figures I presented above are from mid and treble, where it's annoying. The noise from the buffer is close to zero. The noise comes from the input opamps and their resistors.

Good luck with the new board. I hope you apply the tweaks that are audible, practical and make the amp stable, and not those that are there for showing off.

 

[NelsonVandal]

Just to prove it's also measureable, even with a moderate grade soundcard.

I think the difference is greater than 3 dB. The problem is the noise from the soundcard. The worse cross talk in PPAS compared to the other two is probably because of the local feedback = less global feedback. I think I'm going to mod it and remove the local feedback and lower the input resistor value.

I don't mean to back talk PPAS. I think it's a super sounding amp. I use LME49720 and LM6171. From the beginning I used LME49710 in the ground channel, but I found it a bit thin. LM6171 has a warmer mid, but if used in both positions it sounds fuzzy. It's powered by 6 x AAA's and a trickle charger (there's still room for it in a C801). The drawback is low voltage swing since this amp doesn't swing from rail-to-rail.

 

[TzeYang]

are the amplifiers built on a perfboard?

If that's the case, I'm impressed by the benchmark.

 

[NelsonVandal]

Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif are the amplifiers built on a perfboard? If that's the case, I'm impressed by the benchmark.

Of course they're built on perfboard! Except the PPAS. Nothing wrong with perfboard, is there

Maybe if I had an oscilloscope they'd be less impressive, but I doubt some low level oscillations in the MHz range is audible.

 

[amb]

NelsonVandal, your RMAA results' noise figures are improbably good. Are you running these tests with the amp's volume turned down from maximum? Otherwise if the amp under test has any gain at all, the noise should be significantly worse than the loopback. Also, how could the loopback's stereo crosstalk be 10dB worse than with the amp in place? Finally, the figures themselves don't tell enough. The graphs are far more interesting.

 

[NelsonVandal]

Quote:

Originally Posted by amb /img/forum/go_quote.gif NelsonVandal, your RMAA results' noise figures are improbably good. Are you running these tests with the amp's volume turned down from maximum? Otherwise if the amp under test has any gain at all, the noise should be significantly worse than the loopback. Also, how could the loopback's stereo crosstalk be 10dB worse than with the amp in place? Finally, the figures themselves don't tell enough. The graphs are far more interesting.

I run the RMAA tests with the volume at about 12 o'clock, because that's where it's usually at when I'm listening. I know you run your tests at max volume. Can't you please show how you set E-mu PatchMix when you do your tests? I use the default "RMAA", 48 kHz and 24 bit.

Regarding noise, I evaluate all amps by ear using IEM's. The difference in noise is quite audible this way and in harmony with the test results. All my amps have a gain of 4 - 6.

When it comes to crosstalk, don't you think it has to do with output impedance? The tests above were done without load. I can't find my dummy loads. Speaking of crosstalk, I think it's fascinating to see what effect small plugs have. I wouldn't believe it unless I saw it with my own eyes.

I have a high performance Mini3, and I use it as reference. Both of the perfboard amps mentioned above have slightly better crosstalk and lower distortion. The noise level is about the same. It's regardless of load (10, 33, 330R or no load). In both of them there are diamond buffers in closed loop, without multiloop. To my ears multiloop doesn't improve things.

I'll run some test at max volume later. Right now RMAA and/or PatchMix has gone crazy after updating. I really hate PatchMix, it's awful. I'm going to sell the E-mu if I can find some other cheap and good soundcard.

 

[cobaltmute]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif Good luck with the new board. I hope you apply the tweaks that are audible, practical and make the amp stable, and not those that are there for showing off.

I'm going to get onto the board what I am capable of. At a minimum, that means that my board will end up like the orignal PPAS as I can just use the original as a template.

The goal as I see it would be to get the PPA v1 into this size. That was the goal of the original and that's what I'm going to aim at.