RIAA 1k5 LCR

[Andrew]

For more than a few months now I have been working on an inductive phono stage, the design of the RIAA filter is based upon the Tango/Pultec RIAA that was copied by Thorsten and S&B.

Nick's excellent phono uses the S&B's and is loosely based around a topology also devised by Thorsten, who, I think has probably done most to ressurect this kind of RIAA Phono.

The big difference in this implementation is that I have used a 1k5 matched RIAA filter rather than a 600R matched filter as per S&B, Pultec and Tango et al.

Each LCR RIAA modules requires 2 cap, 2 inductors and a bunch of resistors, so quite a bit more to it than the 2 caps and 3 or 4 resistors I have used in the past. Tho' I remember helping James with a real minimalist one that was just 2 caps and 2 resistors.

So, S&B don't make the 600Rs modules anymore. I have since discovered that you can now get 600Rs from Brian Cherry but I wasn't aware of this option when I started. And, having been forced to considered building up my own modules then why not re-evaluate things.....completely.

There could be lots of history behind the 600R choice, but in my opinion its only an obvious reason because 600R matched impedance systems are common in pro audio. However, 600R is quite a struggle to build a circuit around, especially a valve based one.

After much deliberation I went for 1k5 for a number of reasons....

The cap values required are common all garden of-the-shelf-values.

1k5 should be slightly easier to build a circuit around allowing more topologies to be explored - well at least more valve choices.

1k5 is still small enough that the inductors of the required level can be wound accurately, this gets a big problem at 10k a 30H inductor is required.

1k5 is still low enough to be considered a very low Z RIAA and hence should be low noise.

I have since devised a 5k6 RC based RIAA which Simon has been having a go at, I'm really looking forward to hearing Simon's efforts as I think this might be a more cost effctive solution that will get you most of the way there without quite as much hassle. But anyway, I have the first filter completed and it's now under test. [Pics and spreadsheet to follow, Nick is kindly hosting].

cheers,

-- Andrew

[Andrew]

Below is the image of the first module.



And this is the results of the first test run.

http://www.audio-talk.co.uk/images/andr ... e_test.htm

The full xls file for those who have Excel is http://www.audio-talk.co.uk/images/andr ... e_test.xls

[Andrew]

As you can see its a perfect match NOT...

[Andrew]

OK, perhaps I'm being pedantic but the HF is about 150-200mdB up, has anyone else noticed/care?

cheers,

-- Andrew

[simon]

I have since devised a 5k6 RC based RIAA which Simon has been having a go at

And hopefully it'll be working properly soon... I'm really interested to see how it compares to your LCR phono too. I reckon there's about a couple of hundred quids worth of bits in mine at the moment (if bought new) which is pretty good really I reckon.

Simon.

[Andrew]

Bargain, well...almost....definitelty much cheaper than LCR, the S&Bs were £150 each when I last spoke to them.

-- Andrew

[Nick]

OK, perhaps I'm being pedantic but the HF is about 150-200mdB up, has anyone else noticed/care?

Your "out" column seems to all be -ve (other than one), maybe you should normalise that column again?

1k seems to be -0.09, so make that 0, then 20k, is only 0.1db up, seems ok to me. if you look at some of the huge RIAA errors that seem to be measured and dismissed in HFW (in some cases even described as a advantage!!!)

Maybe it would be interesting to see how those numbers stack up with differing signal levels, I think thats the hard thing Dave has had to try and manage, keeping the inductance constant with level.

[Andrew]

Your "out" column seems to all be -ve (other than one), maybe you should normalise that column again?

Good idea...

1k seems to be -0.09, so make that 0, then 20k, is only 0.1db up, seems ok to me. if you look at some of the huge RIAA errors that seem to be measured and dismissed in HFW (in some cases even described as a advantage!!!)

Makes you wonder doesn't it. I suppose a rising HF could be considered a good thing in a dull system, but i would rather fix the dull system.

Maybe it would be interesting to see how those numbers stack up with differing signal levels, I think thats the hard thing Dave has had to try and manage, keeping the inductance constant with level.

Agreed, and a very worthwhile point, it was sort of on my list of things to do once I had established a level I was happy with, but it would be good to know, wouldn't it!

cheers,

-- Andrew

[Andrew]

Well the solid state regulated and buffered HT worked quite well up to the point where a couple of resistors overheated.

One step forward, one back.

cheers,

-- Andrew

[Nick]

Makes you wonder doesn't it. I suppose a rising HF could be considered a good thing in a dull system, but i would rather fix the dull system.

That reminds me, I read a review in Steriophile where a stage that used the 3.18us corner was knocked. I wasn't sure just what the justification for the reason was, it seemed to be based on the fact it would allow more surface noise to be heard. In that case why not just add a 10kHz low pass filter to the end and be done with it.

It was also argued as if the corner was a myth, strange.

Takes all sorts I guess.

Or maybe they dont want to question the fact that $20,000 stages they have reviewed in glowing terms that don't have the extra corner could be sub optimal.

Ahh, no, that would just be my cynical view of such things coming out again.

[Andrew]

If anyone wnat to know what I'm wittering about its this.

Mains TX->5V3->10H->470uF electrolytic cap.

This splits into two off floating LM317 regulators see (roughly) fig 5.36 also fig 5.44 of MJ's 4th Edition. Each of these gives HT +15v this is then fed into a voltage reference to take off 30v to give HT -15 volts and HT +15volts.

These two power HT+- 15v lines are fed into the power rails of two off per LM317 high speed high current op amps configured as voltage buffers to give good PSRR. The input to the non-inverting line of op amps is taken off a simple voltage divider between HT +/-15v to give B+ value and the output B+ is taken directly off the output of the op amp into the valves load, see fig 5.39 MJ 4th Edition. This gives me 4 buffered regulated supply lines capable of sourcing up to 40mA into the valve load.

I succesfully tested about 180v into a 10k on all four loads before I wacked up the variac to full mains and the resistor between MJE340 input and zener started to smoke big time! I think I got too big a mains TX again, why do I always do that?

cheers,

-- Andrew

[Andrew]

Bugger of a job repairing what was left of the track on the PCB, combination of the heat, the layout and de-soldering left fresh air where copper had been, still its only a prototype, might sound crap anyway.

cheers,

-- Andrew

[Nick]

Maybe a 5u4 would loose a bit more voltage for you.

[Andrew]

Hmm, unfortunately not got one, have a 5Y3 or a GZ37, that's it, not very well stocked am I?

cheers,

-- Andrew

[Andrew]

Nearly there, must have lost a LM350T to the smoked resistor on one half but all was OK after I it was swapped out for a new one.

I managed to put 200v through 5k, so 40mA load at around full mains voltage, I'll need to tweak up the voltage set resistors to make a bit closer to ideal as I need a few more milliamps in the real thing.

The resistors that set the current on the zeners and tun into MJE340 are marginal at 3 watt rating as well, 5 would be better.

Next I'll have a rummage around in my box for something like a 3k8 load in high wattage resistors to try the full 30mA per buffer worth of load.

cheers,

-- Andrew