• 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.

50 triode amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Optimizing for lowest distortion I'm looking at the curves for this #10. Right now the powersupply has 400V and can be raised to circa 470V. The anode is at 373V, Ia is 20mA, anode load is 10K. Would it be wise (linearity, aging) to reach for the right most limit at Vg=-60V?

Current situation: ΔV= 531-215= 316 volt, ΔI= 28-12= 16mA (ca. 5VA)
optimal (?): ΔV= 579-215= 364 volt, ΔI= 28-9 = 19mA (ca. 6,9VA)
 

Attachments

  • 50-schema.JPG
    50-schema.JPG
    49.1 KB · Views: 952
  • 10-NationalUnion.JPG
    10-NationalUnion.JPG
    38.5 KB · Views: 929
  • 50.JPG
    50.JPG
    33.1 KB · Views: 887
  • versterkertje.JPG
    versterkertje.JPG
    102 KB · Views: 893
  • 10Y-curves397.gif
    10Y-curves397.gif
    27.4 KB · Views: 168
  • 10Y-curves373V.gif
    10Y-curves373V.gif
    27.4 KB · Views: 873
Member
Joined 2008
Paid Member
The 50 grid's bias is -65V. How much more than that do you want to drive it?

I used 10s as outputs for a few years, driving a horn, so not much power needed. After some experimenting I settled on plate voltages between 250 and 300 and current up to 32mA. Output transformer primaries tried were 16k, 10K, 5K. All sounded very nice. In other words, you have options. Best to try different operating points and see what works for you.

My 46¢ (that's 2¢ adjusted for inflation since 1914 here. It's 66¢ if you want to go back to 1760!)
 
Gain is approximately 65, I have not measured exactly. In the end I'd like to drive a 300B XLS (biassed at 95V) with this front end. The interstage holds at 450Vdc and full signal swing, it's a sturdy build :D

Ahhhhhhhh...

ST envelope 50's are probably the most beautiful tube ever built.

:worship:

urrr... the ST is a #10, the balloon is a #50. Both nice tubes :)

What's the better habit to allow for tube aging? Staying away of the bunching, limiting the Va excursion, or both?
 
Member
Joined 2008
Paid Member
To my knowledge, (value documented above ! : )

As long as you aren't damaging the tube by going outside the various declared limits, the issue with aging is emission. The more you use, the faster the tube ages.

So operating point at moderate voltage and current will give longest life.

Swinging into bunching = distortion.

Also, read the attached paper on thoriated filament life extension.
 

Attachments

  • CPI AB-18 .pdf
    367.9 KB · Views: 98
It was more the effect of tube aging on the circuit at hand I was after but this is an excellent point to consider as well. Indeed, running them less hot prolonges tube life. So does a constant filament voltage. Filament voltage for those bright emitting thoriated tungsten tubes can be reduced it seems. For the dull oxide type it's not recommended... or is this only with the big emitter tubes?
 

Attachments

  • 10Y-starved-curves.jpg
    10Y-starved-curves.jpg
    170.6 KB · Views: 135
Last edited:
Member
Joined 2008
Paid Member
Another Doc for you. Pretty straightforward reading. You probably already know what's below but . . . . .

Recommended drop for thoriated tungsten filaments is 5%. On the 10 , 6.3V is 16% and looking at the stars in Fig. 8 of the CPI doc I'd say that's too low. Notice also that the management program isn't to be started until after 200 hours of use at nominal voltage.

Doing the math given on the CPI doc will tell if the thoriated tungsten filament in question is in the right range for the program.

As far as focusing on tube aging, my personal notion is that if the plate voltage is held constant, on the plate curves the operating point will fall vertically downward in a straight line over the plate voltage shown on the graph and so will the load line. In other words, I believe the load line at the reduced plate current will necessarily be in parallel with the load line at the original operating point. If a CCS is used then operating point and the load line will shift to the right as the tube ages.

And if neither voltage nor current are regulated then operating point will move as determined by resistive drops in the circuit.

'nuther doc. in German. Strangely, Google translate doesn't do a very good job on this one.
 

Attachments

  • Types of Cathodes.pdf
    358.5 KB · Views: 80
  • Regeneration.pdf
    158.9 KB · Views: 65
Extract from the RCA paper:

For turning on filament power [of pure Tungsten filament tubes], a filament starter should be used so as to increase the voltage gradually and to limit the high initial rush of current through the filament. ... Similarly, as an added precaution, the filament power should be turned off gradually to prevent cooling strains in the filament. ... It should be noted that a reduction of 5 per cent in the filament voltage [while maintaining regulation] will approximately double their life.

If this recommendation holds for all Tungsten filament tubes (and not only for big transmitting tubes), Rod has a new task to fulfill: slow turn off.

As recommended by Upton I 'baked' a VT25 for half an hour at 10,0 volt. This restored emission, which was ca 15% back on this tube.

The 200nF coupling capacitor was sufficient but in this test setup changing parts is very easy. The added capacitance does have effect on the weight of bass notes in modern popular music.
I had calculated F-3dB at 4Hz with 470uF cathode bypass on the 27 but 1000uF still made an improvement. Perhaps the reason lays in the age of the old cap, dating from the seventies. Wow, this amp sings. Bass is to die for while it's all silk and honey :D

Next I'll examine decoupling the #50 cathode to AC ground, see what it brings. If I'm right it's custom to set the dominant LF-pole at the front of the amplifier. Why is this?
 

Attachments

  • versterkertje2.JPG
    versterkertje2.JPG
    71.9 KB · Views: 246
  • 50-schema2.JPG
    50-schema2.JPG
    56 KB · Views: 232
Transmitting tube hot cathode characteristics:

After exchanging my easy playing 4 ohm speaker for a nasty 8 ohm bass reflex (with lots of caps and coils going on) reproduction suffered from boom. The lowest register was somewhat out of control. A simple exchange of the 27 and 50 cathode bypass (to HT) made an improvement. Not belonging to the incrowd familiar with Chebyshev's theorem I took the empirical route. Output impedance must have dropped just enough to make a difference. Exchanging this lil tubegiant for the old Marantz PM7200 made no difference in bass reproduction.
 

Attachments

  • hot cathode.JPG
    hot cathode.JPG
    30.5 KB · Views: 161
  • 50-schema3.JPG
    50-schema3.JPG
    53.4 KB · Views: 282
After exchanging my easy playing 4 ohm speaker for a nasty 8 ohm bass reflex (with lots of caps and coils going on) reproduction suffered from boom. The lowest register was somewhat out of control. A simple exchange of the 27 and 50 cathode bypass (to HT) made an improvement. Not belonging to the incrowd familiar with Chebyshev's theorem I took the empirical route. Output impedance must have dropped just enough to make a difference. Exchanging this lil tubegiant for the old Marantz PM7200 made no difference in bass reproduction.
Damping looks on the low side at <2. That since rp is ~1.8K driving an OPT set for 3.5K. Then wdg resistance of the OPT primary & secondary need to be plugged into the DF equation as well.
Many people design for the OPT to be ~3rp, resulting in better DF & lower D%.:)
 
Yeah, I was thinking along that line too. But when I connected the 8Ω speaker to the 4Ω output (offering a higher impedance transformation ratio) there was not much change. It seems the amp is running out of steam at high volume, as most small amps do. One or other speaker makes a difference I noticed. Mind you, differences are marginal but compared to solid state this triode amp has better grip on low notes with the first speaker I tried than with this second. More speakers should be tried before coming to any conclusion.

It passed my mind the opt could be the cause as it's a large 40W GOSS unit, allowing for 130mA. Perhaps the #50 quiescent current is too low for reaching its linear transfer charcteristic. Midrange and high frequencies are good though and primary self inductance should also be constant with quiescent current.
 
Last edited:
If you have a scope & audio signal generator use them, look at the output & sweep the low frequencies. Poor DF will shew the speaker resonance(s) immediately. And you will hear all in a direct way, see where the problems are. A better test than your ears alone. Otherwise you are guessing & flying blind.:)
In your test setup I see many long leads. If they are in the signal path there is a good possibility that parasitic RF oscillations are occurring under some conditions. That will cause very poor audio. The setup looks dangerous as well, very high voltages running in long leads.:eek:If you don't have the test equipment find a friend who does.
 

Attachments

  • John Broskie Comment on Test Equipment.jpg
    John Broskie Comment on Test Equipment.jpg
    177.6 KB · Views: 229
I did do some tests and this is what I got. Input is 0,7Vrms sinus from the signal generator (dropping 1 to 2% with falling frequency, not compensated), output in mVrms taken from the 4Ω tab. First column= frequency, second= 8Ω loudspeaker, third= 4Ω loudspeaker, fourth= 8Ω resistor.

Freq LS-8 LS-4 8res
10 0173 0144 0154
20 1122 1640 0985
30 2645 2795 2401
40 3704 3168 3077
50 4950 3822 3470
60 5990 4640 3710
70 5840 5140 3810
80 5330 5150 3860
90 4920 4920 3847
 
That is good data to work with. I've plotted your measurement results onto a spread sheet, a resonance is clearly visible on the 8Z speaker. The plot & spreadsheet are not really needed but it better illustrates the operating conditions.

If it is in a bass reflex enclosure there may be another resonance just above 100 Hz. In that configuration there are usually two peaks. Are there any crossover networks in the speaker cabinets? Also should do a measurement at mid-band (400 or 1000 Hz) so that a comparison can be made with the LF, especially at the resonance. But looks like both speakers are not well damped. I'll have another look at your data tonight.:) 8R speakers commonly measure ~6.5R on an Ohm Meter.
 

Attachments

  • 50 Amp Response.jpg
    50 Amp Response.jpg
    330.9 KB · Views: 206
Rough Estimate of Damping Factor

First get a rough estimate of the amplifiers internal resistance as seen by the load. The audio generator should be set somewhere near the mid-band of the amplifier, 400 to 1000 Hz, not critical.
Be sure your DVM is still accurate at the frequency set, some older versions are made for power frequencies only on the AC ranges. The spec sheet will tell you.

For your 50 Amp drive, it with a signal that will produce about One volt, again not critical, on the output terminals connected to an 8R resistance load. Measure the resulting voltage. Then disconnect the load resister & measure again.

The internal resistance is simply the change of voltage divided by the change of current in the load resister. Ohms Law.:) By comparing that with the RATED (8R) load, the DF is 8R / Internal Resistance.

If the internal resistance calculated from your measurements was say 3.5R, the DF would be 8/3.5. A well designed triode amp would have a DF of maybe 3. That would assume an OPT of >3 rp. All triode amps without NFB depend very much on the OPT selected.
 
It's a Fluke 87, good for 10KHz at least. Tomorrow I'll rerun tests with a 5K opt, see what it brings. Also, I'm reconsidering the input stage, replacing the Chebyshev filter with an active load. To be continued...

Freq LS8 LS4 8R
100 4180 4620 2823
105 4120 4500 2865
110 4070 4410 3188
115 4025 4430 3210
120 3985 4170 3258
125 4060 4100 3583
700 5010 3748 3615

0,7V @700Hz, no load = 5710mVrms
 
Extract from the RCA paper:
Setting the dominant LF roll off at the front end prevents things like turntable rumble & other such LF from getting into the amp. If they do there will be intermods on the program material. And large speaker cone excursions when we don't want them. On the pair of 6L6GC amps I built around 1960 I stuffed in a 2-pole LF filters on the front end.:)
 

Attachments

  • 25W Amp Front.jpg
    25W Amp Front.jpg
    74.9 KB · Views: 223
  • 25W Amp Schematic 30C.jpg
    25W Amp Schematic 30C.jpg
    201.6 KB · Views: 235
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.