ST70 7199 / 6GH8A amplitude vs frequency

[gadget73]

I know the output on the ST 1700A tweaks a bit. Its not a huge amount, but I've noticed that I can always see a difference in output when I shift the multiplier. I need to change the level pot on that to a 10-turn to compensate for that. Normally I just ignore it, but it does screw with frequency response tests just a bit.


The level set does only work when you're using it to set the distortion levels, but you can use it as a handy way to watch your zero db point. The actual voltage is not critical. The other way to do it is simply measure your output voltages and record them, then convert it to DB relative to your zero point. I use a spreadsheet to get that done for me and to spit it out into a graph format if I feel the need.

 

[Audiovet]

Difference in VRMS between 6GH8A outputs. Tested with a single probe at each output of their .1uf capacitor. There is a tolerance and fluctuation of .001VRMS on my meter

This should ideally be a flatline at 0VRMS from 20hz to 20KHZ

 

[Audiovet]

Again, if I understand correctly ... (and sorry I cannot visit here very often): Difference related to what actual signal at 6GH8 output? I.e. what %? Small differences will be caused because all EL34s are not exactly equal, even in input capacitances. - neither might the wiring be. Some differences might be small, but remember that cathodyne outputs will theoretically be equal only when output impedances (including all influences) are equal. Thus to repeat: What % of output signal does the graph represent?

(As said elsewhere the thd graph is pretty good depending on the output load. I can understand the desire for equality where it should be thus, but one should never lose touch of reality: What will be audible and what not?)

 

[nj pheonix]

I'm at the bottom of your totem pole of advisers here. I did note something in another thread which may or may not come into play as you conduct your tests. I'm not familiar with what you're measuring the input voltage with but some meters are not accurate across different frequencies, that is you might be nuts on at 60Hz but reading off at say 10kHz . If this is known and taken into account already, please forget I mentioned it.

 

[JRUBIN]

OK, I did this three times at 3 different settings on the 331A to ensure it was correct. 1KHZ is my 0 point for measurment
all others are measured as a deviation from the 0db value. I managed to keep input volts RMS to a tollerance of .001VRMS, so pretty good.
The 50ohm inline load made it easier do dial in exact voltages. Something i could not use with my function generator to produce 24W on the output, but ok for 1 watt


Response of 1 WATT from 20hz to 20KHZ

I noticed after the fact that the book goes from 10hz to 40khz. If this is an issue and an important missing piece let me know

 

[JRUBIN]

I'm at the bottom of your totem pole of advisers here. I did note something in another thread which may or may not come into play as you conduct your tests. I'm not familiar with what you're measuring the input voltage with but some meters are not accurate across different frequencies, that is you might be nuts on at 60Hz but reading off at say 10kHz . If this is known and taken into account already, please forget I mentioned it.

You raise a good point. My fluke 25 seems to be a dependable RMS meter with regard to frequency within audio spectrum frequencies. It was tested against the oscope and other RMS testers at various frequency and does well. Ham radio frequencies, on the other hand, well yes, its useless at 160 meters and up

 

[jazbo8]

Perhaps I exaggerated a bit, the generator's output or the input voltage could of course vary and the meter's has its own frequency characteristic as well. But from the above result, it seems your setup is perfectly usable, and shows how flat the amplifier is. If you plot it using log/log scales, it looks even better...

 

[JRUBIN]

So all that is left is matching the resistors to the outputs of the 6GH8A and a final THD test?

 

[gadget73]

Grid resistors on the EL34's being slightly different in value will also mess with things since you'd have a different effective load on the cathode vs the plate. Some amps have you hand match those within 1%, which really is not a bad idea, especially if there is no AC balance pot to make up for it. If yours has a DC balance pot added, then that will also affect the AC balance to some degree, same as off-value grid resistors will.

The fact that it kicks up .1db and runs that fairly flat out past 20khz makes me suspect the output level of the generator picks up at that point. I don't believe there to be any magic circuitry in a stock ST70 front end to account for that. All things considered though, being at worst 0.15db off from flat ain't too shabby. If a bump in tone generator output level above 1 khz can be accounted for, then you're 0.05db off from flat, which is quite good indeed.

 

[crispycircuit]

Does the 6gh8 have more drive? Would it drive KT88s? Of all these style tubes, which would be best to drive KT88s. Scott used 7199 on their big amp, Dyna used 6an8 in Mk3s , what your favorite?

 

[JRUBIN]

Grid resistors on the EL34's being slightly different in value will also mess with things since you'd have a different effective load on the cathode vs the plate. Some amps have you hand match those within 1%, which really is not a bad idea, especially if there is no AC balance pot to make up for it. If yours has a DC balance pot added, then that will also affect the AC balance to some degree, same as off-value grid resistors will.

The fact that it kicks up .1db and runs that fairly flat out past 20khz makes me suspect the output level of the generator picks up at that point. I don't believe there to be any magic circuitry in a stock ST70 front end to account for that. All things considered though, being at worst 0.15db off from flat ain't too shabby. If a bump in tone generator output level above 1 khz can be accounted for, then you're 0.05db off from flat, which is quite good indeed.

I know that the imbalance starts before the EL34's......

Im going to have a look today for some pots/trimpots that meet the following criterial
46K with a 2K upward swing. My pots are old and i don't have many 47K's, so maybe spending the extra money to find some would be a better endeavour

 

[JRUBIN]

Does the 6gh8 have more drive? Would it drive KT88s? Of all these style tubes, which would be best to drive KT88s. Scott used 7199 on their big amp, Dyna used 6an8 in Mk3s , what your favorite?

Favorite is subjective. You can see that in this thread, discussions about "bright sonic tonal fluffyness" and "earthy harmonic depth" and "overly expensive filter caps in the bias that deliver fruity undertones" dont exist, im more in favor with describing an amps characteristics with oscilloscopes and test equipment.

That being said...........

EL Paso Tube amps did a shootout with several 7199's 6U8's and 6GH8A's on the same channel of an ST70 and found that not only do they operate within the exact same parameters, the amount of variation tube to tube (all tubes combined) was such that you could not tell which tube was being used in the application.

When going for a 'component comparison' the closest thing to subjectivity is the Pepsi challenge. which almost all people choose wrong, get lucky, or most importantly, choose the amp with significantly higher distortion, because they got used to the audible artifacts produced by a broken AMP. Guitar amps aside, High Fidelity equipment shouldnt have artifacts, the antithesis of fidelity

 

[jazbo8]

I guess the guy deleted his post... your answer is spot on, unfortunately, we already have too many threads and posts about tube rolling, the "best" capacitor, etc.

 

[JRUBIN]

I guess the guy deleted his post... your answer is spot on, unfortunately, we already have too many threads and posts about tube rolling, the "best" capacitor, etc.

There is no thing in this world without gimmick. Subjectivity has no place in Math. With guitar AMPs its ALL about subjectivity, because they are looking for the right distortion. after obtaining the desired sound you could deduce how it was produced, by the values in math, not the other way around. There is enough snake oil in HI-FI amps for all of us to swim in. The exact difference in sound between a 500$ guitar and a $50,000 is equal difference in which the owner is willing to pay for. Never more, Never less

Ive never seen someone with 10 sets of cryogenically frozen tubes along 10 sets of non cryogenically frozen tubes making a mathematical comparison with any value to it. It just made their wallet lighter

 

[gadget73]

I know that the imbalance starts before the EL34's......

Right, but the EL34 grid resistors are what the 6GH8 "sees" as a load. If they are not matched, it will have an uneven load and you'll get different output plate vs cathode. Same thing would happen if you had a DC balance pot in that position. The effective grid to ground resistance would change depending where that balance pot was set, which would give you an uneven load to work into.

Speaking of which, I probably ought to re-visit the AC balance setting on my Bogen amps. I believe I have meddled with the bias adjustment since I set that. That amplifier uses effectively the same topology as the stock ST70 driver circuit.

 

[JRUBIN]

Right, but the EL34 grid resistors are what the 6GH8 "sees" as a load. If they are not matched, it will have an uneven load and you'll get different output plate vs cathode. Same thing would happen if you had a DC balance pot in that position. The effective grid to ground resistance would change depending where that balance pot was set, which would give you an uneven load to work into.

Speaking of which, I probably ought to re-visit the AC balance setting on my Bogen amps. I believe I have meddled with the bias adjustment since I set that. That amplifier uses effectively the same topology as the stock ST70 driver circuit.

No doubt the grid resistor could be easily checked as well, and should be. ive ordered a pack of 1/2 watt trimpots for the 47K's. In the mean time, i can look at the grid bias resistors as well, they may also be malanigned

 

[jazbo8]

cryogenically frozen tubes

Anyone that spends money on such things clearly has other priorities than science, math, etc. But of course, they have better hearing than most and exquiste taste in "hi-fi", so what do we know...

 

[JRUBIN]

We's spoken about the fact that some work had been done on the left channel before the testing started, giving not an entirely accurate BEFORE for the AFTER. Note that as I wait for the POT to arrive for the left channel and still have to test the grid resistors of the EL34 afterwards that the right channel has had no changes done to it with the exception of 1 resistor so broken that there would have been no way to retrieve any results.

Im still critical about the output of the phase inverters, most of all, as those small changes will have a much larger effect through the EL34's. That being said, as the grid resistors are before the EL34's they too share a high degree of relevance

The grid resistors for their part can be matched 1%, but will still see different voltages due to the lack of my 47k pot which im waiting on.



Even after ALL of this is done...... and testing is good, my homebrew PAS3 is in need of work and represents a separate project

 

[JRUBIN]

EL34 grid resistors
LEFT RIGHT
F 940 940
R 955 974


Incidentally in answer to another question

7199 Phase inverted outputs
Rear PIN1
FRONT PIN8

 

[gadget73]

Are those the grid to bias supply resistors, or the bias supply to grid resistors? The ones going to the bias supply are what the phase inverter sees as the load. Not sure what they used on those, but anything in the 100k-330k range would be my guess. Its possible these live on the board, I've never worked on an ST-70 so I don't know where they may have hidden this.