Pentode Phono

Post by **dave slagle** » Sat Feb 06, 2010 9:22 pm

JJ and I have been looking at this for a while and the goal is a Cheap phono that uses RC correction that kicks buttocks.

One of the goal is to try to do the whole thing with shunt filtering and somehow by tweaking things I came up with this.

It sims nicely in the frequency domain but the thing that bothered me was the load the pentode sees gets vanishingly low at high frequency. The loss of gain at the large load is actually part of the correction.

I had a back and fourth with stephie and i figured it would go better here.

shinebox · Post by **shinebox** » Sun Feb 07, 2010 10:49 pm

Hiya,
Funny you mention this - I was just helping my cabinet maker buddy put together a very simple phono stage using... D3a,
Very interesting! May I ask why only shunt filtering? Less evil resistance in the filter path and strong drive for the output stage?

I don't have a good feel for how optimal loading of pentodes should work - nor how a harmonic profile which is a function of frequency will sound. Then again, extra distortion at HF may not be a huge issue if the harmonics are ultrasonic...

Will have to sim it and see how much current the tube is going to swing at HF. My one concern is ticks/pops may be a bit problematic (see Morgan's phono stage chapter on these exciting HF cartridge resonance)

Either way, delighted to see we can benefit from Stephie's wisdom again.

Cheers

sbench · Post by **sbench** » Mon Feb 08, 2010 4:06 am

Pops and ticks will be more pronounced for certain. The problem is partly gain distribution. At 10kHz, Dave's circuit, as shown, runs out of drive somewhere between 100-200mV at 10kHz applied to the inverse RIAA input. That's reasonable HF headroom, but not spectacular. You can gain back more HF headroom by running the pentode at higher current... not higher power, just higher current.

Don't forget, from a distortion view, pentode's love low load impedances.

On me being back. You know how hard it is to shut up a broad, right?

-Stephie

shinebox · Post by **shinebox** » Mon Feb 08, 2010 11:10 am

Hey Stephie,

Keep talking all you want!

I hear you regarding the current... I must keep remind myself how good a vertical loadline can look with a pentode at blistering currents.

When I get a chance, I may try and sort out a sim with E55L (I think the US # is 8233)... that should be happy running at 50mA, for a gm of about 50-55. Have to reinstall Spice and all the library stuff first though as having some problems with it...

Cheers

Post by **dave slagle** » Tue Feb 09, 2010 4:42 am

OK peoples.

I'd like a bit more current and bias on the 5687 (or whatever tube ends up there) so maybe another 50-100V of B+ will be needed. I was hoping to get away with a Genrad 1203 to make things simple and keep the PS out of it for some (ie me) hence the 300V.

We are actually looking at the russian 6E5P as the pentode. I used the D3A because it was a model I was familiar with.

I also still find the idea of the variable gain of the pentode with load as part of the filter really interesting.

Last thought is I wonder how much the iriaa having 15dB of gain from 20Khz -100Khz is making things look worse than they really are. (I don't have my head around this one yet)

dave

shinebox · Post by **shinebox** » Tue Feb 09, 2010 9:51 am

Hi all,
Gotta say, the more I think about the pentode/shunt scheme, the more I like the idea.

I have very little time this week to play, but just wanted to throw in some ideas for consideration:

1) Tube choice
Input could also be E280F, and maybe worth investigating the Russian 6J52P... apparently a bit like their version of the D3A. Haven't had time to investigate fully but there's a thread on diyaudio and a warning about dodgy sellers...

Then could save the 6E5p for the output stage.

2) Speaking of the output stage...
Could consider running a 5687 (or 6E5P) long tail pair on the output stage and avoid the 47uF cathode cap, stacked supply etc. Would give you polarity switching etc.

3) If we want to get really crazy... I have to check the phasing of this, but it could be that the other grid of said long tail pair might provide a nice point to inject (frequency equalised) B+ rail noise sample into the output, which would probably require removing the PS decoupling between stages. Inspired by Broskie's stuff that.

4) Speaking of noise, it wonder if might be better to return some of the RIAA shunt to B+ instead of ground. Again, haven't had time to think about this properly and do some sims.

In any case, getting very excited about the possibilities here. Wouldn't surprise me if this thing killed the LCR - how's that for fighting talk!

Hope to have more time to spend on this soon...

sbench · Post by **sbench** » Tue Feb 09, 2010 4:49 pm

Dave
Post the .ASC file. The DC is creaping due to clip at HF. Please annotate the nodes, it helps figure out what you're measuring. Here's a slightly altered spice file. Note that R11 is a lot higher.

Stephie

shinebox · Post by **shinebox** » Wed Feb 10, 2010 9:22 am

Hey JJ,
You mentioned causing trouble... I like to do that too.

That should certainly work; I'd be curious about the noise at these levels, and actual dynamic shunt impedance of the VR tube may be comparable to that of the small cap at HF and affect the EQ... just guessing here. That's why I though a long tail pair stage at the output could be interesting, though I suspect your scheme has the potential for better sound. I won't go into level shifters and servos for the same reasons... esp. given the target audience for the design.

I gave Clips a shout last night and we diagnosed some of the issues above - the dc for example was ameliorated by running the sim longer - probably the time constant on the output. Dunno why the dc initial conditions biz in Spice didn't do the trick... also must get my installation fixed so I can try and replicate/play

sbench · Post by **sbench** » Wed Feb 10, 2010 4:02 pm

For the initial conditions to work right, do a DC analysis, noting where the operating point IS, then set the voltages and currents with an .IC statement. Then you don't need to worry.

On the VR Tube. Getting the bias right is going to be very tricky. As the D3a bias point shifts (if only by volts, which is normal) the bias on the second stage will change substantially, which is likely to alter the sound. Think of it this way... we're asking for a high gain circuit. If the DC gain is also high (direct coupling and fixed biasing) then the bias points are also subjected to that same high gain. Transformers (I can't use the abbreviation any more, since that means something altogether different to me now- giggle) and / or capacitors avoid that issue.

-Stephie

reVintage · Post by **reVintage** » Wed Mar 24, 2010 10:14 am

Hey Dave,

Don´t know why I haven´t seen this thread. Low-budget is fun!

To make it even simpler you could use the russian triode/pentode 6F12P. Both halves with S=20mA/V and the triode with mu=100.

The one, with lost the same type of EQ, I´ve been working on for quite a while, is full of sand with a "gyrator"-loaded DHT 3A5(Oh, so cheap) on top and a pair of 2SK170 at the bottom. No second stage needed(gain=40dB) as signal is taken at the lowZ output of the "gyrator". Finally I´ve ordered the PCB´s so it will not be long until it plays.

About cracks and pops this is not a problem IRL. Have used the cascoded version with E88CC(B+=300V) of this for a very long time together with AN IO and AN S-7 former.

izzy wizzy · Post by **izzy wizzy** » Thu Mar 25, 2010 9:07 am

That's why I though a long tail pair stage at the output could be interesting, though I suspect your scheme has the potential for better sound.

And like Lars, I have no idea how I missed this thread. The comment from Shines is very relevant for me as that is what I run at present. See attached. I did it as I only had a 5mA LL1660 tx hanging about and wanted to hear what it went like getting rid of the parafeed output stage.

I've been messing around with a shunt version in SPICE as well and have a version which I thought worked very well. But reading Stephie's comments about overload, bias shift etc may mean it isn't so good after all. Think I need to look/learn a bit more.

reVintage · Post by **reVintage** » Thu Mar 25, 2010 9:49 am

Hey Stephen,

Great thread this, and the LTP looks nice.

But I thought the mission here was a Cheap phono

.

About your circuit I think you might miss the point a little as you have added a series resistor, R16. I understand it though, as loading is tuff.

My theory, after having lived with this kind of EQ for many years, is that the lack of the above is what makes it play so great. There might be drawbacks with high loading at the highest frequencies but it is after all a compromise. And it plays music.

Will go home at my lunchhour and post a single pentode hybrid RIAA that I sketched on yesterday. Unfortunately gain will not reach 40dB.

izzy wizzy · Post by **izzy wizzy** » Thu Mar 25, 2010 10:07 am

Hi Lars,

I got a bit carried away when I saw a phono thread and the LTP comment and posted what I was currently using

From scratch, I agree mine won't be cheap but it was cheap for me as the output tx was just lying around on the shelf

I've often thought if I got rid of it, I could have a second cheap phono lying around for funs.

Attached is the shunt phono I've been messing with. Not sure I fully understand pentodes which is why I was a bit apprehensive posting my sim as the biasing looks a bit strange to me but the AC part seems to work well at low level. I have to look at Stephie's comments to see how they look in this circuit but not sure just yet how to look at bias shift, HF AC overload and the like in the sim. Lots more to learn

reVintage · Post by **reVintage** » Thu Mar 25, 2010 10:34 am

Hey Stephen,

About pentodes (or cascodes) there are no secrets about the way they work:

Zout is ideally infinite.

Gain is defined as Gm*Rload, the same for AC or DC. Lets say the anode resistor is 10k and Gm=10mA/V, gain is 100=40dB. As Zout isn´t infinite IRL, Ri must be parallelled with Rload when calculating gain.

To linearize and add some local feedback the cathoderestor can be left unbypassed to some degree. At the same time this makes Ri higher.

Some of our friends here with deeper insight, surely will have a lot more to tell.

reVintage · Post by **reVintage** » Thu Mar 25, 2010 1:43 pm

Back to the "El Cheapo"-concept.

If one can accept some sand on top and 34dB gain we can do with only the pentode.

I am not so sure the load should be so heavy, so in this one it is around 5kohm at 1kHz. So now we don´t have so close to a short at 20kHz.

Ah, why don´t go for 40dB as below.....

reVintage · Post by **reVintage** » Thu Mar 25, 2010 2:29 pm

Ooops,
Got a little carried away. The only thing needed to make the 34 go 40dB was to short out the cathode circuit. Indications of ca 3 times higher TDH. Overload characteristics will be worse too.

reVintage · Post by **reVintage** » Sun Mar 28, 2010 11:13 am

Hey guys,

Did some simple THD-sims of the 36dB circuit above.

With the rev-RIAA removed and all at 1kHz I checked with:

51k load ie the loading at 20Hz=56dB gain: THD=1%, input level 100mV

4,6k//51k load ie the loading at 1kHz 36dB gain: THD=1%, input level 800mV

420ohm load ie the loading at 20kHz=16dB gain: THD=1%, input level 300mV

All levelsare peakr.

Have anyonre checked the unequalized max-levels at different frequencies from a typical PU?

izzy wizzy · Post by **izzy wizzy** » Tue Mar 30, 2010 9:43 pm

Hi Lars,

I didn't have an IRF610 model handy so subbed an IRF510. That might have led to a problemette. I ended up with a lot of current down the valve; about 45mA and so about 12W dissipation. Is that what you get?

I also messed about with my para EQ circuit and found it very sensitive to G2 volts. As such, with the values I've chosen, in particular the cap values (not willing to go larger) I think it a non-starter as it would be very susceptible to valve variation and therefore probably ageing.

reVintage · Post by **reVintage** » Wed Mar 31, 2010 7:50 am

Hey Stephen,

You forgot to add the DN2540 model(get it from Supertex), adjust Ua to 150V. Get IRF610 model from IRF, but IRF510 should be equally good when simmed, but not IRL.

No problem whatsoever to get a stable Us. A 150V stab tube (150C4) fed via a CCS is one example.

izzy wizzy · Post by **izzy wizzy** » Sun Apr 04, 2010 12:47 pm

Hi Lars,

The models that I used are fine but I was using the first circuit you posted and couldn't get that to bias up at all. The second one was fine. With the IRF510, I changed the EQ parallel Z to 39k for good response. All food for thought

cheers,

Stephen

Post by **dave slagle** » Thu May 06, 2010 5:05 pm

got one built...

izzy wizzy · Post by **izzy wizzy** » Thu May 06, 2010 7:22 pm

Went looking through some schemos on the pooter, found Jim Hagerman's SKP (Simple Killer Phono) which uses an E280F in pentode and shunt EQ. Not sure if the circuit is on the public domain though. Bit more traditional WRT coupling and biasing.

That's quite a departure from you trad build style. How's that feel?

Look forward hearing the reception you guys get with it.

cheers,

Stephen

reVintage · Post by **reVintage** » Fri May 07, 2010 1:27 pm

Hey Dave,
Nice to see you got the act together. And how does it play?

Post by **dave slagle** » Sat May 08, 2010 7:30 pm

it sounds good. I don't have anything to compare it to at the moment but next weekend at the LSAF i'll put it up against Jeffrey's LCR and the Artemis labs PH-1 with teflon caps and nickel outputs.

dave

IslandPink · Post by **IslandPink** » Wed Nov 17, 2010 6:37 pm

Any news on this comparison , Dave ??