• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

To vary a signal pentode to triode in stages

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
too late edit for above:
The resistor pull-up for the Mosfet drain, instead of the CCS pull-up, may restrict the max screen current in UL mode though, unless the Mosfet is heavily loaded with idle current. (as the screen V rises in UL mode, the pull-up Mosfet drain resistor to B+ has less voltage across it.) I guess it just boils down to which Mosfet gets hot, the follower with a drain resistor or the CCS Fet with no resistor.

edit: the CCS will be the cooler way I think, since the drain resistor requires excessive drain current when the voltage drop across it is high.

Acck!!:
I just realized that the drain resistor method does not allow correct current thru the coupling cap to the plate. The pull-up CCS is required after all.
 
Last edited:
smoking amp

I am trying to follow...:confused: I picture is worth a thousand words;)

The way I am trying to plot the curves is with a fixed DC Screen voltage representing what the Quiescent screen voltage would be under operating conditions. In this case the plate will go below the Screen voltage the same way the curves are traced for the Pentode mode at a fixed Screen voltage.

I tried something different today. It works but I probably need a better non electrolytic capacitor for better operation. It must withstand the large plate voltage swing and block the DC coming up from the plate. I used a small 120V variac.

The additional screen current definitely affects the way the traces look, specially at lower plate voltages. This is a 6L6GC at 250V DC screen voltage.

Variac to Pentode Side
An externally hosted image should be here but it was not working when we last tested it.


Variac around 40%
An externally hosted image should be here but it was not working when we last tested it.


Variac on Triode Side
An externally hosted image should be here but it was not working when we last tested it.


Variac on Triode side with Screen Voltage supply off. Same volts/div for x and y as previous ones.
An externally hosted image should be here but it was not working when we last tested it.
 

Attachments

  • Fullscreen capture 4232012 125756 PM.bmp.jpg
    Fullscreen capture 4232012 125756 PM.bmp.jpg
    177.8 KB · Views: 156
3CKAh.jpg


I will save the UL feedback until the morning reading now ( thanks to all , noticed the lovely curves ) .

On a lesser note . I was a bit hasty over the RC snubbers and have implemented the 2 capacitor one resistor solution recommended in the text mentioned previously . One thing that is plaguing my tests is Ethernet over the mains ( I have to remember to switch them off and how many and where ) . The suggestion to fit ferrite beads seems a good one ( keep everything close to the diode ) . If anybody is building anything and would be willing to listen to these options I would be fascinated to know what they think . The consensus is don't fit the single capacitor option often recommended by testing houses ( said to result in harshness and better not fitted at all ) . However the two capacitor and one resistor option works well and would more than meet emission regulations ( I am 95% sure the RC network would ) . The unsuppressed version is a marginal failure , if the equipment is well screened it might not be . I recommend carbon composition resistors and ceramic / film caps . If carbon comp nearest value will do fine . These values are the ones recommended in the text . My application only sees 54VDC . If tempted to see if the neon tube has anything to offer please scale the resistor to suit the voltage . The idea is these tubes are very fast and might mop up something if it rises to striking potential . In this application one electrode is cold . I suspect what I saw was capacitance . The part No might be obsolete . I tried another neon tube of unknown origin , it worked fine . I was very surprised how much power these neon diodes could absorb . I tried one briefly without and limiting resistor . It got mighty hot yet did not fail . That suggests it might have some unexpected qualities . Hope this is useful ? Always make sure the probes are tight when doing measuring tests ( and snapped on , warnings about high voltages and isolation as one would imagine ) . It would be interesting if by adding a series resistance one might better approximate a thermionic diode if doing this ? Useful if so to have in reserve .
 
re: Avincenty

Your Variac technique looks good. Certainly getting decent curves.

-------------

Mosfet version:

The current sense resistor in your diagram is placed so that it automatically sums plate plus screen current correctly, so the addition of a Mosfet follower simply needs to avoid polluting the current thru that sense resistor. The plate V resistive divider can be placed just like the Variac to the top of the sense R. The Mosfet drain would be pulled up by a CCS to a B+ or something staying above the plate V.

A cap from the Mosfet drain to plate, to transfer the screen current over, may be optional in your tracer setup, since it is already looking at cathode current where both are summed. If the cap is put in, then you need the CCS pull-up on the drain (to avoid variable drain load resistor current added to the plate current). If not connected though (the cap), then I think you can just use a load resistor to B+ for the Mosfet ( then the B+ supply neg. returns to the top of the sense R, so screen current goes thru it).
[ so either AC screen current either goes across the cap and back around thru the tracer supply ,and thru the current sense R; or for the no cap variant, DC current just goes thru the separate B+ supply and back around thru the current sense R. The second uncoupled variant captures the total AC + DC screen current in the current sense R, and is easier to do and more useful for the tracer purposes. The cap coupled/CCS version is more useful if you want to make a UL trioded device for an amplifier.]

A pull-down resistor on the Mosfet source ( which is connected to the screen) may be optional. If it is used, it's current draw must avoid polluting the current sense R for the tracer. So it would go to the top of the sense R, like the divider.

The Mosfet gate gets a cap from the R divider, and a high value R to a screen DC bias supply.

In any case, you just need to follow the current loop around for any support devices, to avoid polluting the tracer's current sense R.
 
Last edited:
Thanks Smoking-Amp

I did some simulations in LTSPICE IV. I followed what you said but I think the easiest one is the one I have attached as it does not require additional B+ supply to keep the Drain above the gate voltage. Although the plate voltage is varying it will always be a tad higher than the gate as the gate voltage is the output of the divider and going all the way to triode is invalid as it will yield a boosted triode with much higher G2 than the plate. Also, even-though the Screen supply is in series with the mosfet source, the current stills goes via the drain to the sense resistor via the transformer centertap.

Do you think this will work in circuit?
 

Attachments

  • ulwithmosfetdiagram.jpg
    ulwithmosfetdiagram.jpg
    107.5 KB · Views: 95
  • ulwithmosfet.jpg
    ulwithmosfet.jpg
    313.4 KB · Views: 93
Hi Alfredo . That looks nice and simple , thank you . I've only came back to valves recently and find it very stimulating to think through . I hope someone will follow this up and say what happens in real examples . I was recently diagnosed with Parkinson's so only switch on my soldering iron when convinced I can get it to work . I hope others don't mind doing the exploring for me . I sort of fancy using Gu 50 and my EF 184 . The idea was to use this arrangement with the EF 184 .
 
Nigel, sorry to hear about your diagnosis. Hope you can manage.

This is a hobby for me and I have been learning a lot since I started a couple of years ago. It is therapeutic for me.

I went to the shop (garage) and hooked it up.

Works really well:D Very clean traces this time. I had a 900V enhanced N Mosfet laying around. I think I am going to wire it in a box permanently so I can use it easily as an attachment to the curve tracer. It should also work for local parallel feedback testing. The cool thing would be if I could get it to readout the percentage on the dividing network.
 

Attachments

  • 12%25209%253A24%253A30%2520AM.jpg
    12%25209%253A24%253A30%2520AM.jpg
    68.3 KB · Views: 57
  • 12%25209%253A25%253A18%2520AM.jpg
    12%25209%253A25%253A18%2520AM.jpg
    85.9 KB · Views: 39
  • 12%25209%253A26%253A31%2520AM.jpg
    12%25209%253A26%253A31%2520AM.jpg
    93.9 KB · Views: 39
Last edited:
Very nice solution, I like the simplicity the floating screen supply allows!
And the curves look great! I will add this to my tracer too.

Next addition to consider for the tracer is a way to drive both the g1 and g2 grids with the step generator using g2/g1 Mu scaling. The g2 needs higher voltage steps than usually available (HV amplifier may be called for), plus some current drive capability. The g1 steps get divided down by g2/g1 Mu so that both grids do the same amount of work in controlling the plate current. g1 can work in either the enhancement mode like g2 (ie, positive) or it can use the usual negative bias range. (so DC biasing adjustments for each grid is called for tracing or use)
For a typical TV Sweep tube with internal Mu of 4, this boils down to g2 operating at twice the usual grid1 swing, and g1 operating at 1/2 the usual g1 swing. (so g2/g1 steps 4X)

Your curves clearly reveal the continuity of triode to pentode as a varying Mu triode.

-----------

Nigel, sorry to hear of your diagnosis too. I suspect a fair percentage of the participants on the forum are battling some disabilities or age related problems. Myself included. A good hobby to relax with.
 
Last edited:
It's not so bad except nothing stays still when I solder . This is what got me going in the link below ( p 12 especially ) . I've read so many accounts of UL before , somehow this one said it is the lowest distortion point . Mostly everyone says a good compromise . Because this never is used with input valves maybe something very important has been overlooked ? Smoking amp said ideal into resistive load . We should be able to get close to that .

I was with my Pro Audio chums last night . They say valves are a box of crayons .I said nonsense . I am trying to change that was my reply . They say if I do I will not have a valve sound . I say nonsense again I will have space around what I hear . We might soon be able to say for certain .

I subscribe to the view that if an amplifier has less than 0.1 % distortion and good harmonic qualities with 5 Hz - 3dB to 40 kHz - 1 dB then no one on the planet can hear any crayons in use . For my personal use 1% ( 3 % on a good day ) . I do find HF rolll off is very dependent on how it came about as to whether I can hear it . FM is not obviously lacking . I was acquainted with Michael Gerzon who was noted in the subject of filter design . I never talked to him about it however followed up his work . Fascinating reading if others want to follow it . Michael had the mathematical ability to destroy doubters . Echo's in filter systems etc . That's why I beleive UL is not easy .

P12 and your curves do look to be telling the same story .

http://www.pmillett.com/file_downloads/ultralinear.pdf


I once asked Michael what he thought about negative feedback . He said he never had but if he had he would have said this . " Some amplifiers need a lot and some very little " he then burst into laughter and said "you can bet on one thing , nearly all have the wrong amount " . Some say Michael was above Einstein in his abilities , sadly no longer with us . Good digital is probably mostly Michael 's work ( loss-less compression ) .
 
Very nice solution, I like the simplicity the floating screen supply allows!
And the curves look great! I will add this to my tracer too.

Next addition to consider for the tracer is a way to drive both the g1 and g2 grids with the step generator using g2/g1 Mu scaling. The g2 needs higher voltage steps than usually available (HV amplifier may be called for), plus some current drive capability. The g1 steps get divided down by g2/g1 Mu so that both grids do the same amount of work in controlling the plate current. g1 can work in either the enhancement mode like g2 (ie, positive) or it can use the usual negative bias range. (so DC biasing adjustments for each grid is called for tracing or use)
For a typical TV Sweep tube with internal Mu of 4, this boils down to g2 operating at twice the usual grid1 swing, and g1 operating at 1/2 the usual g1 swing. (so g2/g1 steps 4X)

Thanks Smoking-Amp

I don't follow the g2/g1 proposal. What kind of curves will that give, kind of like a "balanced" triode curve? Any links for me to read? . When you say steps on G2, are these always positive steps from 0V up or similar to the G1 that go negative.

Alfredo


Alfredo
 
The g2/g1 Mu scaling idea came up back in (I think) Pete's Red Board (Engineer's amp) thread as a way to improve on screen drive. It reduces the required screen drive voltage swing (and current) by half, so is much safer for the health of the Sweep tube's screen grid. Interesting curves as well. The screen current can also be combined with the plate current there via a Mosfet follower/driver drain too.

The g2 steps are always positive, although some pure screen drive setups go negative on g2 during cutoff.

In any case, the hardware required to do the g2/g1 Mu scaling on the tracer will also allow pure screen grid drive testing as well. (you just add an R divider for the g1 drive add on, if it's in the negative bias no current range.)
 
Last edited:
Nigel, this is all very interesting, and is actually picking up where it was left off many months ago in another thread.

Notice on page 154 of the UL manual the chart showing different amounts of UL feedback depending on the tube.

One outcome of this may be an easy approach to measuring where that point is for each tube in question.

I've got a set of 25% and 40% transformers I had wound to experiment with using 6P1P tubes, that I need to dig out and test.

Here is the thread you might want to look at if you haven't already.

http://www.diyaudio.com/forums/tubes-valves/97384-adjustable-distributed-load-discussion.html
 
I'm glad I reactivated it Gimp . I recently built a 4 transistor amp in the fashion of HC Lin . It measures in a way some would doubt possible . I never finished it . Somehow I didn't like the way it sounded ( it was very detailed ) . I restated my interest in valves and realized only two valves required ..

I am very lazy about maths ( I am not without a gift in that ) . However friends around me are very gifted so I ask them .

I made good progress with diodes , use cheap ones and 3 component snubbers . The values I lifted from a worked example seem not very transformer dependent . Thanks for pointers .

Nigel
 
It's stuff I love to eat . Thank you so much for that . I was totally locked up physically . I now have the minimum dose of dopamine and it works ( it took time ) . My friends seem to worry more than me . I can now go to the bar and bring drinks back ( 90% gets back ) . That's good enough for me . Dry solder joints are a problem , lead free doesn't help .
 
BTW . I was talking with a freind who has a very sensitive ear if a good idea comes along . I said we should make a TR 34 ( EL 34 without the extra grids ) . I said my guess is to move the g1 up 20% towards the anode as we had the space . He has access to one of the Russian makers . Do you think it would be of value ? I did say even just removing the spare grids would help . This is mostly a marketing thing . However I thought it up and doubtless will never see a penny . I did say if he does it I want it sold at normal EL 34 prices . I have no idea who he knows as the company , I forgot to ask .
 
Clarification to the tracing method

I had been going over what I have done with my curve tracing setup and realized that since I do not have the voltage swing that the transformer provides that I need to create it myself to emulate the transformer.

On the drawing, the fixed UL DC voltage (V1) must be properly selected depending on target B+ and UL%.

V1 = (B+) - (B+) x UL% to set the lower limit of the Screen Volts when the Plate voltage is to reach 0 Volts on the plate drive signal.

In the 6550 example below, using a B+ of 450 and a 40% tap I would need to setup my tracer V1 to:

V1 = (450V) - (450Vx40%) = 450V - 180V = 270V

If my rectified Plate drive in my curve tracer is set to max at 700V, then the aggregate screen voltage will go from 270V(0v plate) to 550V(270 + 700x40%), reflecting the same screen voltage swing as in the 6550 example below.
 

Attachments

  • ulwithmosfetdiagram.jpg
    ulwithmosfetdiagram.jpg
    107.5 KB · Views: 129
  • 6550ul.gif
    6550ul.gif
    26.4 KB · Views: 125
Alfredo . It gets very exciting following this . Do you have a small pentode anywhere you could try ( EF 86 ) ? I wrote to my freind in Slovakia about this hoping he will . We have to do it through Google translate because neither speaks the others language . I read back the translation and it says the reverse of what I meant . I said I have started a thread on UL feedback in small signal valves which ideally feeds a resistive load as the best I could get it to translate . Thank you so much for doing this and thank you smoking amp for waking up my interest in maths . 200 page tome on UL , that made my day . No it made my week .
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.