Output stage and SE->Bal implementation

[Pars]

I'm getting ready to build up one of Colin's tooleaudio I/V stages to put into my Rotel RCD-991. This is a dual PCM-63P player with pcm100 filter. I have currently been listening to it modded with THS4031s for the I/V stage and AD8599s doing the buffer/bal conversion.

stock output stage:

U302 is the single I/V opamp, U306 is the buffer/conversion opamp (dual) and U304 is an anti-aliasing opamp (socketed now, but not populated.

I am trying to decide whether to

1. Retain the AD8599s as buffers/conversion
2. Replace the AD8599s with different opamps
3. Use transformers? If so what?
4. Other?

I want to retain the muting relay functionality. The markup below is a first stab at what I think needs to be removed (green) as well as the output (to I/V stage) and input (from I/V stage) back into the player. DC coupled, so the coupling caps will be removed and jumpered. Any suggestions welcome.

Edited December 10, 2020 by Pars

[luvdunhill]

I like the ADs you have in there. I have a slight preference for the LME stuff these days.

All that to say though, I think everyone should try a transformer at least once. So, I'd do that. I might be persuaded to let you borrow a set of transformers, if you'd like. I have plans for them, but it's down the lie a bit so I'd be willing to let you try and see if they work for you.

[Pars]

Marc,

I might take you up on that (transformers), for a listen. In order to get a sense of how much these would be should I like them so much I felt the need to buy some, what are they? Sowter?

Filburt suggested replacing the ad8599s with ada4898-2s also.

[luvdunhill]

Yup, 9545 with the wire upgrade. The are made for IV use in current out DACs. They are 1+1:5+5.

[Pars]

Ooops, guess I misunderstood you. The transformers are intended for I/V use? I assumed they were just for SE->bal conversion. I don't think I even want to try transformer I/V.

[luvdunhill]

I don't think I even want to try transformer I/V.

Wimp! what do you have to loose? You could chat with Doug, but they are bifilar transformers, I sure you could cram them in somehow

[Pars]

Actually there is quite a bit of room in there, though looking at them I'm not sure how you are supposed to mount them. I think I'll remain a wimp, though ya never know...

BTW, is an I/V resistor the only other thing you need with these, or do you have to AC couple them as well?

[luvdunhill]

you got it. Just a resistor and the transformer!

Edited January 18, 2011 by luvdunhill

[jcx]

I don't do "how does X op amp sound" but technically speaking the ADA4898 is something new and uniquely suited to the objective, technical demands of I/V conversion

it can tolerate 4 mV of I/V DAC switching "glitch" spikes on the input without moving the input stage into nonlinear operation (-120 dB incremental linearity change with 4mV diff input)

this isn't obvious from the data sheet but Scott Wurcer and another AD insider I know revealed that it uses the Gilbert Multi-Tanh linearization scheme - with a patented noise reducing twist

it also is faster, has rather higher output current capability than the NE5534 it replaces in this circuit - which reduces feeedthu with the highly desirable feedback C establishing the 1st pole of your anti-alias filter

the feedback C at the I/V op amp cuts slew rate and glitch feedthru to following circuits - if the I/V op amp is fast enough and has enough current capacity to prevent the glitches from simply "shooting" thru the C at high frequency

you can see that the circuit designers were concerned about the switching glitches causing nonlinearity with the conventional bjt diff pair input of the NE5534 by the inclusion of the input filter to the I/V chip

I would try the ADA4898 with feedback C and bias the output 2-5x the PCM-63 output current, skipping the input filter

[Pars]

Thanks, but if you read the thread completely, the ADA4898 will be used for buffering / SE->balanced conversion only. I/V will be done with a discrete stage designed by Colin Toole (cetoole here).

Regarding the input filter: yes, I just noticed that when I posted this thread... not sure why I didn't see it (and remove it) when I replaced the NE5534s with THS4031s (with much bypassing).

Edited January 18, 2011 by Pars

[dsavitsk]

Assuming there is no DC across the primaries, you probably just want some standard 600 ohm line level transformers. There are lots of these available for not too much money. My favorites are probably the Cinemag CMOQ-4H which are quad filar on 80% Ni cores. Only issue is that ordering from Cinemag takes some time (I have a couple dozen around, though made to my specs -- send me an email).

The Lundahl LL1527 is a good reasonably inexpensive option (it dispenses with some of the features that make Lundahls expensive and which you don't need here), though if the output level is high the cores might be too small.

Jensen has a dizzying array of choices, but the Cinemags are probably better and cost less.

Probably not a bad idea to pick up some Edcors which can be had for $10 each.

Hammond also makes some like the 145E (available at Mouser) which look pretty crappy, but might be OK to play around with.

Oh, and eBay is a good source for vintage stuff, something like an Altec 15356, though frankly I don't get the obsession with vintage transformers -- I think new is good here due to improvements in core material.