ST70 7199 / 6GH8A amplitude vs frequency

[JRUBIN]

One small matter I find missing (hope I have not missed relevant commentary!), is the action of the 82pF phase capacitor over the 270K pentode load resistor (on my schematic drawing, in series with 18K). With the ST70 this plus other Cs start to diminish loop gain upward of some 7kHz, and thus NFB, ergo: A rise in distortion. This is what I see reflected in the distortion graphs.

First I did a calculation with 18K and 82pf. I saw a cutoff frequency of 100K so i re-re-read your comment

Ok so there is the 7K.......... However..................... See below

My 270K's in the schematic are actually 1% matched ............ at 306K!!!!

I think I might have found a problem here

 

[JRUBIN]

The 1% resistors that were 306K were replaced with ones that were 274K 1% as the closest match I had to 270K as per the schematic


1st chart from before at 306K using the iphone test input

2nd chart at 274K using the same test. Almost no difference at all. I was sure that this would have brought about a positive change....
distortion still starts launching at 1000hz and the numbers are almost identical and right back where i started. Unless someone knows the significance of a component that causes this to happen at 1K im at a loss.....

 

[gadget73]

Those numbers really are not bad for what it is. Its a tube amp, they just do not measure amazingly well. Not sure what your output wattage is exactly, but it seems reasonable based on the measurements Dave got from his round of testing with an ST-70.


Single Channel -- @ 20 Hz = 39.8W RMS. @ 1 db down, THD = 1.0%
@ 1 kHz = 45.6W RMS @ 1 db down, THD = .18%
@ 20 kHz = 43.5W RMS @ 1 db down, THD = 2.3%

Both Channels -- @ 20 Hz = 31.6W RMS @ 1 db down, THD = 1.9%
Operating @ 1 kHz = 37.5W RMS @ 1 db down, THD = .32%
@ 20 kHz = 34.1W RMS @ 1 db down, THD = 3.3%


I've intermittantly fooled around with measurements on different amps, and they all tend to increase THD with frequency. It also usually climbs at the bottom end, but thats partly a function of the transformer limitations. If you measure something with small iron, you'll see the THD figures go through the roof at some magic point, which will be where the transformer saturates.

 

[JRUBIN]

Swapping the 6GH8A for a 7199 shows the same characteristics

Short of anyone finding fault with this, im going to make the baseline off of this for right channel testing.
When left and right channel match, I will go back to 24 watt testing.

Test from just before

Same test swapped 6GH8A with 7199 (in fairness the 7199 is an ooooold tube from the amp)

the last thing i can think of is that slight imbalance on the phase inverter on the 47K 1% resistor
that increases in difference as frequency goes up. I think I need a pot in there

 

[JRUBIN]

Those numbers really are not bad for what it is. Its a tube amp, they just do not measure amazingly well. Not sure what your output wattage is exactly, but it seems reasonable based on the measurements Dave got from his round of testing with an ST-70.


Single Channel -- @ 20 Hz = 39.8W RMS. @ 1 db down, THD = 1.0%
@ 1 kHz = 45.6W RMS @ 1 db down, THD = .18%
@ 20 kHz = 43.5W RMS @ 1 db down, THD = 2.3%

Both Channels -- @ 20 Hz = 31.6W RMS @ 1 db down, THD = 1.9%
Operating @ 1 kHz = 37.5W RMS @ 1 db down, THD = .32%
@ 20 kHz = 34.1W RMS @ 1 db down, THD = 3.3%


I've intermittantly fooled around with measurements on different amps, and they all tend to increase THD with frequency. It also usually climbs at the bottom end, but thats partly a function of the transformer limitations. If you measure something with small iron, you'll see the THD figures go through the roof at some magic point, which will be where the transformer saturates.

The Iphone test is only in and around 4-5 watts RMS output

 

[dcgillespie]

You might find it instructive to test the amplifier -- PROPERLY LOADED -- with the one lead of the 18K or 82 pF temporarily lifted. The distortion will improve at high frequencies, but at the expense of stability, so be ready to shut it down if any instability is noted. The distortion will likely NOT level out however, as the the HF response of small signal pentodes employing a large plate load resistor is notoriously poor, and becomes part of the HF stability scheme. Therefore, OLG will still decrease at high frequencies (reducing NFB), but not as much as when the step network is in place.

Dave

 

[JRUBIN]

Having matched all of the components correctly. The only outstanding issue is the 47K resistor. Im not going to make adjustments to the feedback if I still have a known issue to deal with .
Here are some earlier results Ive posted, I want to revisit this and chart the offsets just like ive done for the other frequencies. Im going to make a guess beforehand and
say that at 1000hz and under this difference becomes comparatively insignificant.... but well see what the results say


OLD chart for review....... outputs of 6GH8A


HZ tube VRMS OFFSET
500 F 5.97
500 R 6.00 +.03

1000F 5.96
1000R 5.99 +.03

5000F 5.85
5000R 5.91 +.06

10000F 5.66
10000R 5.77 +.11

15000F 5.37
15000R 5.56 +.19

20000F 5.13
20000R 5.33 +.20

 

[JRUBIN]

Here are the results as compared against the previous chart. There is a definite similarity. Its time to get a POT in there and see if this can be balanced
Im getting the feeling that the higher the frequency and a constant offset of resistance, the greater the difference. below 1000hz its negligible.



THD% chart

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

 

[gadget73]

which side is higher, cathode or plate? Vaguely curious what mechanism is in play here.

Side note, I set the inverter in my Bogen by summing the two channels and adjusting the pot till it was flat. I re-visited it after it was together using the THD meter. Where that gave me best THD performance was not quite where the signals were equal according to the scope. Not sure if it was a loading thing or making up for imperfections in the transformer or output tubes, but you may find that theoretically ideal and actually ideal are not the same.

 

[JRUBIN]

which side is higher, cathode or plate? Vaguely curious what mechanism is in play here.

Side note, I set the inverter in my Bogen by summing the two channels and adjusting the pot till it was flat. I re-visited it after it was together using the THD meter. Where that gave me best THD performance was not quite where the signals were equal according to the scope. Not sure if it was a loading thing or making up for imperfections in the transformer or output tubes, but you may find that theoretically ideal and actually ideal are not the same.

I hadnt actually checked which component of the tube corresponds with the front and rear EL34, but ill let you know.

Using 10K for example , the back output was .18VRMS > than the front

Also I havent made any correlation between the 6gh8a outputs and the values for the EL34 product

Understanding this correlation would shed light on the totality of the offset in values shown. but all i know is that
zero would be a good starting point

 

[jazbo8]

Please re-read dgillespie and Audiovet's posts, you seem to be jumping the gun with the THD vs frequency measurements... Could you please provide the frequency response (at 1W output into 8R dummy load), distortion at several output power levels (1kHz) and the 5kHz square wave response?

 

[JRUBIN]

Please re-read dgillespie and Audiovet's posts, you seem to be jumping the gun with the THD vs frequency measurements... Could you please provide the frequency response (at 1W output into 8R dummy load), distortion at several output power levels (1kHz) and the 5kHz square wave response?

After seeing that the phase inverted outputs of the 6GH8A are unbalanced, im curious as to why this issue shouldn't be corrected first. Also, If I dont take THD measurements now, I wont have a comparison 'before' for my 'after'. If im off base here, let me know.

Also as far as frequency response, are you looking for a particular measurement. Ive been using VRMS and converting to watts. An earlier post indicated that the input signal voltage varies with frequency, making matching difficult

I didnt entirely understand your request. You lost me at '1 watt output' as a setup criteria and then at several power levels' However, from what I gathered 5khz at 1 watt into an 8 ohm dummy load

 

[jazbo8]

Also, If I dont take THD measurements now, I wont have a comparison 'before' for my 'after'.

We are trying to get a baseline of the whole amplifier before the changes (although you have done some already).

Also as far as frequency response, are you looking for a particular measurement. Ive been using VRMS and converting to watts. An earlier post indicated that the input signal voltage varies with frequency, making matching difficult. I didnt entirely understand your request. You lost me at '1 watt output' as a setup criteria and then at several power levels' However, from what I gathered 5khz at 1 watt into an 8 ohm dummy load

The frequency response is usually taken at low output level or under "small-signal" conditions, so simply set the input level with the output at 1W/1kHz, then sweep through the frequencies - leaving the input level fixed. The units are usually dB vs. freq., not W vs. freq. The THD readings at several output levels, give you an idea of the clipping point, you only need to do it for 1kHz.

Please take a look at the Stereophile measurements for the ST-70 II or any other amplifiers, to get an idea of the typical tests on power amplifier if the above is not clear to you... BTW, your square wave looks pretty good.

 

[JRUBIN]

We are trying to get a baseline of the whole amplifier before the changes (although you have done some already).


The frequency response is usually taken at low output level or under "small-signal" conditions, so simply set the input level with the output at 1W/1kHz, then sweep through the frequencies - leaving the input level fixed. The units are usually dB vs. freq., not W vs. freq. The THD readings at several output levels, give you an idea of the clipping point, you only need to do it for 1kHz.

Please take a look at the Stereophile measurements for the ST-70 II or any other amplifiers, to get an idea of the typical tests on power amplifier if the above is not clear to you... BTW, your square wave looks pretty good.

Let me know if this would work, so that I may try it this evening:


Can I set the output to match a particular DB value on the analogue meter of the 331A (say at 1khz), voltmeter, not distortion, and measure all of the other frequencies against that, given a stable input voltage.
-4db sits in the middle of the display. -2 would be ideal, if the expected swing did not deviate more than 4db. (id imagine a half power, 3db increase would be considered substantial, 4 would be nuts)

i know 1 watt would sit at 2.82VRMS which would be either too high or too low for the scale depending on the set range , at either just under +1db or just under -10db. If we are expecting less than 1db swings then yes 1 watt would be super

Id have to monitor the input voltage from the source (iphone) to make adjustments on frequency changed to ensure the input stays the same. I know that at 1watt and less im no way near full power so that may help things

picture used as example and not for any displayed reading on it

 

[gadget73]

You can do it that way, but personally I'd get it as close to 1 watt as you can, and then use the sensitivity control to set the meter on the HP to the 0db mark. I've done it that way on the ST1700A. The HP's meter makes it a little easier since there is room above the 0db point. The ST puts 0db at the end of the scale, so any rise in output just runs off the scale and tells you nothing.

 

[JRUBIN]

You can do it that way, but personally I'd get it as close to 1 watt as you can, and then use the sensitivity control to set the meter on the HP to the 0db mark. I've done it that way on the ST1700A. The HP's meter makes it a little easier since there is room above the 0db point. The ST puts 0db at the end of the scale, so any rise in output just runs off the scale and tells you nothing.

Sensitivity only works with setlevel.... not voltmeter, however they are analogous so, i can do the test at setlevel as well since were not measuring voltage on the display.

Tests at low wattage are much easier for me to do as the dummy loads can be cooled with ambient air, indefinitely. i will use your method and see if i can adjust with the vernier for 0db. You see my consern with going off scale and ruining the test, but then
it doesnt really matter because the output is so small, i can simply retest, just want to avoid it

 

[jazbo8]

Can I set the output to match a particular DB value on the analogue meter of the 331A (say at 1khz), voltmeter, not distortion, and measure all of the other frequencies against that, given a stable input voltage.

Yes, that's the idea, set the volt meter reading with 1kHz, so 2.83V means "0 dB", and measure all the other frequencies relative to it - do not touch the input level while you are taking the measurements. The frequency response should match the ST-70's spec very closely.

 

[JRUBIN]

Yes, that's the idea, set the volt meter reading with 1kHz, so 2.83V means "0 dB", and measure all the other frequencies relative to it - do not touch the input level while you are taking the measurements. The frequency response should match the ST-70's spec very closely.

Ill have to put a meter on the input jack as well to ensure that the input voltage doesnt fluctuate with the frequency changes.

 

[jazbo8]

Ill have to put a meter on the input jack as well to ensure that the input voltage doesnt fluctuate with the frequency changes.

Since the input impedance does not change, why would the input voltage change?

 

[dcgillespie]

Unless an oscillator of very good design is used, it is not uncommon for the output of typical test oscillators to vary slightly over different bands or frequency settings. The Heath IG-5218 (for example) certainly exhibits this characteristic, while otherwise being a very good general purpose oscillator.

Dave