FireFly Single Ended Tetrode

[TheGimp]

I'm using wing 'C' (Svetlana) 6P1P-EV tubes. I think this one was marked as 1974 production date, but I'm not sure. I can check to see. I've got a smatttering fo 1973 thorugh 1984 or so.

Thanks Metallo, hopefully Doug does not mind if I use the name as well. The AX84 site shows a copyright on the amp design, but not the name so hopefully he won't. Besides, this is not a commercial product, just another amp for fun.

I could change the name to "Fire-Fly" or maby "LightningBug", but the name just doesn't seem to have the same ring.

rotaspec, I only ran it about 10 minutes at that power level and I doubt it would last long with that level of abuse. I'll get back to abusing it again in a couple of months when the OPTs come in.

Mean time I'm going to try to figure out if I can use a computer sound card to drive the input and measure the THD. I'm hoping the THD of hte soundcard is << than the amp, so I should be able to get reliable measurements.

I'll be using a Soundblaster Audigy since my M-Audiophile 192 is dead and I still haven't received a response from the company about a RMA number to send it back for repare.

[TheGimp]

I got around to starting to take distortion measurements this weekend. I say "starting to" as our son got home from Germany Friday night, after being away for nearly 5 years in the Air Force.

So, (1) I didn't have much time to play with the amp, and (2) I'm still reading the manual on the HP333A distortion analyzer so readings are going slow and the first three attempts were way off.

I did run a couple of plots on AudioTester V2.2, and compared the calculated distortion with the HP333A.

I’m running with -6dB GFB, and measured the distortion with respect to plate current using the screen regulator to control plate current.

Input drive was adjusted for 800mVrms output into 8 ohms (80mW) in part because I didn’t’ want to overdrive the sound card I’m running with AudioTester V2.2.

Distortion goes up rapidly above 0.125W.

All dB measurements are minus, but the –sign is left off because it is used as a space setter for the table.

…………………….......25mA----35mA-----45mA---50mA
Screen Voltage….181-------205--------226------236
Vgk………………......10.42-----10.42-----10.56----10.59
THD-HP333A……..1.14%----0.92%---0.42%----0.32%
THD-Calculated….1.126%--0.71%----0.36%----0.287%
Fund……………….....3----------3----------3----------3
2nd Harmonic……..42--------46--------52---------54
3rd Harmonic……..63--------65--------67---------68
4th Harmonic……..95--------90--------92---------90
5th Harmonic………82--------100------92---------95
6th Harmonic………91--------95--------105-------98
7th Harmonic………85--------82--------84--------86
8th Harmonic………104------86---------92--------94
9th Harmonic………96--------91--------86--------84
10th Harmonic…….95-------93---------93-------95

Not really surprising results. Distortion decreases with cathode current. The more current the higher we are on the bias curve and the farther away from the compressed lower traces for higher negative bias conditions. However it looks like most odd harmonics are suppressed more with higher bias levels as well. This is more noticeable when watching the FFT while adjusting the screen voltage (and hence the plate current).

More current is better, but in particular below 35mA distortion really starts to go up.

[TheGimp]

I took a quick look at distortion on the output of the triode driver this morning before comming to work. With the cathode set for 35mA as previous, the screen drive shows 0.5% thd on the HP333A. I was expecting most of the distortion to have come from the output, but it looks like I need to work on the driver as well.

One thing I'm going to try on the output tube is to replace the parallel strings of LEDs with a PNP and a single string of LEDs from base to collector, then a 470 Ohm resistor from base to emitter. This configures the transistor as an emitter follower with the LEDs as the voltage reference. It should yield a lower impedance than just the LEDs with the balancing resistors in each string. I may try a bypass cap on the LEDs as well.

[TheGimp]

I tried changing the bias on the input 6N1P and it made little difference in the distortion. I suspect I'm at the limits of my power supply to swing through the plate curves. Lower distortion will require a higher B+ which will allow operation above the compressed curves.

Tonight I took several distortion measurements of the amp with different levels of NFB at full power (2.88W, 4.8VRMS 8 ohms just before clipping).

FB_______None__-6.75___-7.85___-8.9____–26.8
THD_____12%____6%____5.9%___5.4%___4.6%

And comparing bias schemes:
LED/EF___Resistor___Resistor/Bypass___LEDs
5.8%_____30%______5.9%____________5.8%

LED/EF - 5 LG M67K with 2N3638A Emitter Follower and 390R bias to LEDs.
Resistor - 266 Ohms
Resistor/Bypass - 266Ohms with 470uF bypass.
LEDs - three banks of cheap red leds, 5 per string, with 2R7 per string.

Feedback up to 6dB really makes a big difference in distortion. Beyond 6% it has a lot less measurable effect.

Just an unbypassed cathode resistor really produces distortion for some reason. I even dropped the input drive slightly to see if It was from clipping and it made little difference.

The remaining three bias mechanisms all had about the same level of distortion.

Next I think I'll set the feedback to -6dB and try listening tests with all four cathode bias schemes.

[TheGimp]

Current Schematic:



Screen adjusted to 220V with:
Vak = 256V
Vgk=-10.18V
to achieve Ik=35mA

V3 was placed in the grid circuit of the output tube to make the screen and grid voltages come out right for the plate current at 35mA. Something wrong with the 6P1P model no doubt.

[rotaspec]

Originally Posted by TheGimp

Just an unbypassed cathode resistor really produces distortion for some reason. I even dropped the input drive slightly to see if It was from clipping and it made little difference.

I read somewhere that an unbypassed cathode resistor means a dramatic change in the internal resistance of the tube compared to bypassed (but I can't remember which way it goes).

I have played with the bias on 6N1P until I am blue in the face, and I haven't found a plate current or load that I am completely happy with. Even without any real test equipment, I can tell by listening that there is some distortion there. A 6N3P seems to sound better in the same circuit (but it is a complete socket rewire job, so comparisons take some time). At least the mu is similar (36 vs 33), so NFB changes are probably not required. I would love to see spectra of the two of them under similar operating conditions.

[rotaspec]

Originally Posted by TheGimp

I'm using wing 'C' (Svetlana) 6P1P-EV tubes. I think this one was marked as 1974 production date, but I'm not sure. I can check to see. I've got a smatttering fo 1973 thorugh 1984 or so.

How do I decipher the date code on these?
Mine have 8908 just below the 6N1N-EB (don't have cyrillic letters on my keyboard so that's as close as I can get).
They also have a rectangle with some cyrillic letters and 0490 .r, and also a diamond with 25 in it.
Is that 8th week 1989 production, and inspection of 4th week 1990?
They were not boxed specimens so I have no other data.

Gary

[TheGimp]

"Effect of Cathode Capacitor on P-P Output Stage" by Robert M. Mitchell shows it worse with the bypass cap than without if no GNFB is applied. With GNFB there is a cross over point where clipping is approached, beyond this distortion is worse for a bypassed cathode resistor.

I thought I had another article with similar testing but I can't find it.

Unfortuantly I don'at have any 6N3Ps or I'd do a comparison with the spectrem analizer and distortion analizer. I've got 6N2P and 6N6Ps if you would like any data I can get from them.

As far as date codes go, the tubes I have only have one 4 digit number so it is pretty clear which two are year and which two are week/month/lot or whatever. That is consistent with what I've seen on ebay at least.

I would think the numbers below the tube designation (8908) would be the date code and would represent 8th week of 1989.

I've been thinking about the distortion from the 6N1P, and wonder if it might be possible to reduce distortion by shifting the bias point but keeping the current source set point where it is. I tried changing the current from the current source but did not change the cathode bias.

[TheGimp]

I checked some date codes tonight. They are all below the tube designation which is below the wing C symbol. Typical date codes were 0286, 0985, XII 80, 0985.

I optimized the distortion of the input stage and got it down to half of what it was originally(near 1%). I'm loading the anode with a current source and the Cathode has an LED and 100R for the GNFB return point. The final numbers for the conditions for the 6N1P-EV were:

19.4Vp-p at the plate (to drive the output to 4.8Vrms into an 8 ohm load)
Va-k = 124V
Ik= 5.35mA
0.42%thd

Too bad it made very little change in the output distortion which is dominated by bias current effects at 35mA. I expect that in a PP amp or an amp with less distortion in the output stage, it would make an appreciable difference.

Distortion for the amp with about 7dB of GNFB stands at 5.2% at onset of clipping (4.8Vrms into 8 ohms).
At 2.8Vrms out (1W) it drops to 3%
At 1Vrms (0.125W) it drops to 0.95%

Soundwise, the wife is quite pleased with it, and wanted to know if I could build one and a set of speakers for her car!!!

Uh, sorry but no.

[dovla]

Originally Posted by rotaspec

How do I decipher the date code on these?
Mine have 8908 just below the 6N1N-EB (don't have cyrillic letters on my keyboard so that's as close as I can get).
They also have a rectangle with some cyrillic letters and 0490 .r, and also a diamond with 25 in it.
Is that 8th week 1989 production, and inspection of 4th week 1990?
They were not boxed specimens so I have no other data.

Gary

Production dates on Soviet tubes are always month & year, sometimes reversed and sometimes with Roman number for month so 0985, 8509 and IX 85 would all be September 1985. Any stamps (OTK 'diamond', rectangle with numbers, etc) are QA (and/or military) inspection marks and can also contain dates but usually don't.