PAS visits the naughty side of town (red lit)

[jackinnj]

PAS visists the naughty side of town (red lit)

So, I said, what the heck, I don't have any 5842's, EC8010 or WE417's hanging around, why not use Morgan Jones preamp avec LED and pop an inexpensive LED instead of the cathode resistor in my PAS modded amp.

Here are the trace identifications -- The GREEN trace is the first stage 300k/1.8k per the Vorhis mod. The BLUE trace substitutes a cheap read LED for the 1.8K -- very little current is flowing through the LED.

The RED trace substitutes 157K for the plate resistor and utilizes the LED while the MAGENTA trace just used the 1.8K.

I used a very ancient LED --

[IMGHTTPDEAD]http://www.tech-diy.com/RIAA/PAS_RedLit.gif[/IMGHTTPDEAD]

 

[EC8010]

Well, I don't know, seventy views and no replies? Could we have a diagram of what you were playing with? Oh, and by the way, I'm fining you one Audio Precision for not including a diagram.

 

[SY]

What??? You don't have the Vorhis PAS mod committed to memory? For shame!

Trust me, you'd hate it.

 

[ArtG]

Sy,
Hate it? Why?? Are you referring to a particular article of the three that he published? Admittedly, the RIAA equalization that he devised, was inaccurate by later standards, but nevertheless a fair improvement over the original. While there are certainly better circuits around today, I thought that the mods were well worthwhile considering the overall simplicity and the time period.

 

[SY]

I went through various iterations of PAS mods back in the '70s and '80s and finally came to the conclusion that unless you were willing to do a gut-and-rebuild, it was polishing a turd. Inadequate open loop gain, poor ability to drive the network at high frequencies, coupled with horrible overload properties dooms this sort of two ECC83 circuit to mediocrity. Adding a cathode follower to drive the network helps greatly (basically turning it into an SP-3/6), but it still never gets to, errr, "blameless" performance.

It's better than stock, but that's setting the bar pretty low.

 

[andyjevans]

Well, I built the Last PAS a very long while ago and I was entranced by its performance. Vorhis worked a long time on this and I felt he knew what he was doing. I'd be curious to build it again with better parts - I have very fond memories of it.

 

[jackinnj]

Vorhis made an "apparent" error in the RIAA circuit, and it doesn't appear to meet the results which were published in the article. There does not seem to be enough gain (the feedback loop impedance is too low) at DC.

Using the LED reduces the THD% dramatically. This is shown in Morgan Jones "Valve Amplifiers" -- and the reasons explained therein.

But getting back to the matter at hand -- the LED is used on the cathodes of a 6922 -- there is plenty of gain for the first section to be drive a passive shelving network. The second half uses an "anode follower" described by Titschler in AudioXpress, October 2003.

I've tried to get hold of the Author, I believe he was a student at Wright State when he wrote the article, his company seems to have disappeared, however. His boards are a bit big to squeeze into a PAS.

Edit: Apparently the DIYAUDIO web-nanny doesn't like you to use words which appear on the front part of a young lady's anatomy --

 

[jackinnj]

Just some notes -- THD% is about 0.55% -- here's a graph of the RIAA error --

[IMGDEAD]http://www.tech-diy.com/RIAA/6922_Graph.gif[/IMGDEAD]

 

[SY]

Edit: Apparently the DIYAUDIO web-nanny doesn't like you to use words which appear on the front part of a young lady's anatomy --

We really need to whip our server monkeys to do a better job there. "T!tschler" should fool the little ******s.

 

[jackinnj]

The little lady that could -- is that a gun in your pocket or a 7868?

[IMGDEAD]http://www.tech-diy.com/RIAA/6922_RIAA.jpg[/IMGDEAD]

 

[jackinnj]

I tweaked one of the resistors and the RIAA error averages about 0.2dB over the audio band. An infrasonic filter is probably a must.

Using the Agilent Red LED in both cathodes of the 6922 resulted in about 0.25 dB of gain vs the 200 ohm resistor. I bypassed the LED's with 1,000 uF -- was in there to begin with. Distortion is lower by a hair as well, but I don't want to take the PCB's apart to measure it without the RIAA components.

This Tritschler RIAA amp is a bit different from that which appeared in AX, i.e. 5842 in the AX article, vs 6922. The design is very finicky about component values, however.

 

[EC8010]

Was your distortion curve measured via a pre-RIAA driving the RIAA stage?

 

[jackinnj]

I just ran the distortion at one frequency, 1kHz -- and without the feedback loop -- just a standard triode into a Wima 1u/400V polypropylene with a 1 meg metal film as load, about 10mA.

 

[EC8010]

But your first graph shows a percentage against frequency, I thought that was distortion? Was it perhaps a typo, and you meant dB?

 

[SY]

Huh, I took it to mean % error from RIAA.

edit: Never mind, wrong graph. I'd better get that coffee...

 

[jackinnj]

But your first graph shows a percentage against frequency, I thought that was distortion? Was it perhaps a typo, and you meant dB?

Oh, that (top) graph was with an ECC83/12AX7 -- a first stab -- the assembled RIAA section uses a 6922 with a 22k/5W plate resistor -- but I made one test at 1kHz without the RIAA hooked in. (It's a passive RIAA network).

 

[jackinnj]

Here's a graph of the RIAA error for the 6922 passive RIAA with 200R cathode resistor and an Agilent Red LED -- you could shift the graphs up or down so that the 0 intercept is at 1kHz.

The charts are pretty similar to the simulated Morgan Jones passive RIAA.

[IMGDEAD]http://www.tech-diy.com/RIAA/Comparison.gif[/IMGDEAD]