PGA2311 and CS3310 Performance vs Input Impedance

Status
Not open for further replies.
I'm a hifi collector/magpie/tinkerer and have a couple of Primare units with these. One is a P30 with the CS3310 and the other an SPA21 which uses the PGA2311. Neither of them sound good to me. I've spent over a year at this point mucking about with them, swapping other components from my collection in and out and scratching my head. I've also had them both professionally recapped and restored to a fairly high standard. At this point I'm confident that the issue is with the two preamps.

I've gotten into the schematics and modelled in Multisim to try to see what's occurring. Both units have very nice simple preamps when run in bypass mode and the only thing which they have in common and in my opinion could possibly be causing the issue is the volume chips. What both configurations also have in common apart from using volume chips is that those chips are being run outside of specification.

I'm going to focus on the SPA as that's the one which has really had the time and shows the most egregious problem. The PGA2311 is not buffered on the input and is seeing an impedance of around 3.3k. The datasheet recommends 600ohm or less for optimal performance and no more than 2k. Subjectively (but extensively subjectively tested) the upper mid range is harsh and forward to the point where many tracks are unpleasant to listen to for a period of time and the bass seems a little woolly.

I'll mention the P30 as an aside because it's not had the same focus in testing and these things can be subjective, but it may contain a clue about both chips. I find it to have muddy bass in comparison to other pres and a lack of presence in the upper mid and treble. I notice in the spec sheet it talks about not exceeding the input impedance spec due to interactions with its capacitance and wonder whether the PGA has the same issue and becomes non linear outside of spec.

So I guess my questions are: Has anybody tried running one of these chips (especially the PGA2311) with an unbuffered, higher than spec impedance and if so did you find the same? Even better has anyone run tests and has hard data they could share?

I must stress that while I'm confident of the issue with the SPA / PGA2311 the P30 / CS3310 is not as severe and could be subjective, so my focus is on the former.

N.B. In the blurb for one of the SPA21s successors (SPA22 or SPA23) they talk about needing to buffer the PGA2311 and having learned their lesson from the SPA21. Given that the P30 sounds basically ok, just not perfect imo, I wonder if they thought that they got away with it with the CS3310 so could with the PGA as well and it didn't work out because it's more sensitive to it. Or maybe it's all in my head. Comments / test results appreciated.
 
Inspection of the datasheet reveals the PGA2311 is fabricated in CMOS, which implies the output opamp is going to be a CMOS one. My experience with CMOS opamps is that they tend to have particularly poor PSRR. In order to mitigate this weakness if I have to employ a CMOS opamp (such as internal to a DAC chip) I pay particular attention to the output loading. The lighter the load, the better the PSRR.
 
Do you have the details of the PGA's power supplies? If its fed from the usual 3 terminal regs they tend to have fairly high output impedance when only trickling current into a handful of opamps. Shunts give much lower Zout so you might wish to explore that avenue.
 
Can we assume that a PGA2311 is harmless when immediately following a PCM5102A DAC?
Is it better to insert a 20 kHz 600 ohm passive lowpass filter in between? First order? Second order?
How to generate the high quality + 5 Volt - 5 Volt supply from a low quality USB + 5 Volt? A Murata NKA0505 DC / DC converter? Followed by a passive LC filtering or some gyrator filtering, eating less than 200 mV DC?
 
Hi steph - any kind of DC/DC converter is going to need a passive (LC) filter after it. You really need to know the switching frequency - if its around 150kHz then I use a 150uH series L. If its running at 1.5MHz as some of the newer devices do, you can get away with a smaller inductor, 33uH I'd suggest.
 
Well this is interesting. Many thanks to braxalito who has found some fairly serious issues with the configuration of the rest of the circuit so the PGA2311 in itself might not be to blame. It looks like there's some work to do fixing the other issues before the PGA can be tested subjectively again.
 
Status
Not open for further replies.