DIYHiFi.org

Duplex
Whatever...

The intention was to make a discrete pre, with passive RIAA and should be two stages (no more than that), single-ended.
It was important that it had sufficient gain (many commercial designs have too low output), and it would be nice that it was an improvement over what I was using before (OPA637, biased to class-A), or I would drop it.

It is (sounds) better, and that's why you see this thread and this schematic.
The last files I posted were still not right.
The problem was the high-ish value of the series resistor (R5) I was using, which with the input capacitance of the last stage caused premature HF rolloff. I then tried to compensate for this on the wrong place - the RIAA cap that sets HF rolloff (C3). It didn't sound natural (!). Duh...
It was when lowered the value of the series resistor (R5) that the problem was no more, and I now have an accurate RIAA curve, without any issues.
I once cascoded the first stage, and then the second stage, and either way it didn't sound good. High impedances are an issue, one advantage creates another disadvantage, and then I would have to buffer both stages - it would get away from the main intention for a simple, minimalist circuit.
This pre was started with the 2SK170 jfets.
I could not get with the 170s the gain I have now with the 2SK369 jfets.
Also, to get some gain the drain resistors were much higher with the 2SK170, wich was a problem, as described above.

Seeing all the factors, I think I made a good compromise here.
Good sounding, simple, easy to build.
But the components must be these, respect them and it will work very well.

I did my best to make a clear schematic, here it is.
PS: There are no kits - don't ask.

The simulated RIAA curve, taking into account the output impedance of the first stage (important!).
The RIAA curve is accurate to around +/- 1db, I'm not going to loose my sleep with that.

Carlos,
why you decided to put R5 there?
the output impedance of the first stage is R3-1.2K, and RIAA network can be recalculated for this impedance.
I beleive you don't need this extra atenuation and higher noise contribution from the second stage.
I would like to add also input resistance switch 33K 39K 47K 62K 75K
And I would like 75us C3 to be in parallel with R8.

On the simulator it's not just 1.2k, it's a little more.
The series resistor isolates the first gainstage from the capacitance of the RIAA network, I would not like to drive that directly.



Yes, that would be nice.



Mmm... didn't think about that.

EDIT: forgot to say Dimitri, I was not able to get the RIAA curve right with lower than 3.3k series resistor (R5). I lost some time with that but then I thought I was already suffering from 'simmulitis aguditis'.

Hi Carlos,

Some constructive criticism I hope.

Haven't had a solid look at the cct but the RIAA response looks to be a bit light on in the bass. I would like to see considerably more at 50Hz and a rolloff starting around 25Hz, to ensure a solid and tighter bass with less sensitivity to room modes and resonances/warps etc..

Ideally a 50KHz break would also brighten the highs a little and track the lathe eq. better.

Cheers,
greg

Hi Carlos,

Some constructive criticism I hope.

Haven't had a solid look at the cct but the RIAA response looks to be a bit light on in the bass. I would like to see considerably more at 50Hz and a rolloff starting around 25Hz, to ensure a solid and tighter bass with less sensitivity to room modes and resonances/warps etc..

Ideally a 50KHz break would also brighten the highs a little and track the lathe eq. better.

Cheers,
greg

I can't get more bass without affecting the midband. Probably it's my simulator that is no good (free stuff), or it's not easy to get that with passive EQ.
I don't feel there's lack of bass...

Dimitri, the simulator treats the input jfet as if it had ~1.6k output impedance (with 1.2k drain resistor, as on the schematic).

I'm getting a decent RIAA curve on the simm with a 1k series resistor (R5).
Dimitri is right, this will make less noise, and will allow to use lower gain on the output stage.
The schematic will change then, soon.
Need to test this.

Carlos,

If you have it built, why not measure it? At this point, who cares what a computer simulation says it should be doing. More important to measure the actual circuit to see what the truth is.

Easiest way to check this quickly is using an inverse RIAA filter. Copy the one from Audio Amatuer or from my website. Then use a 500Hz or 1kHz square wave. Any imperfection is immediately noticable.



Maybe I complain about this too much, but simulations seem to be way overused these days.

jh

I'll second all of those opinions.

BTW.......I know someone who still has 2 inverse networks left over from an estate sale..........hint, hint.

Jocko

Allpassive correction sound good, but have problem with higher noise...Better is to separate it at two stairs and between both to give next amplifying stage. By this configuration you give much more better SNR and sound stay the same...

Carlos, I made some quick sims, and yes, Guru was right, you don't have enough gain at LF. To have 40dB gain @ 1hHz you should have initially 60dB of dc (LF) gain. This in not the case for your amp, so the active load is required to boost LF with extra 12-15dB.

First six components and VCVS is the inverse RIAA circuit.
Here are the amp input (green trace) and output (blue trace) signals.
Try to think also in this direction:
http://www.stevens-billington.co.uk/riaapage.htm

YGM.

Thanks.