Page 8 of 8

#106

Posted: Wed Jun 18, 2014 12:52 pm
by Alex Kitic
Thank you, it's a very interesting information, and a text I have not read before.

If the color is 2000K, it must be quite yellow - and it is, but how yellow is yellow? The glow of the filament is indeed yellow, not at all white. But the reflected light looks more similar to "warm white" CFL, which is why I thought it is 2700K.

Even more interesting is the influence of heater voltage on temperature, where almost 10% increases just 100K - this difference is something the human eye cannot distinguish, but the intensity of reflected light is what we base our estimate on.

Further, at 2000K expected life is much longer than 1000 hours, as for instance declared for GM70, but it does increase further with 1900K - at the expense of a little emitting power.

That said, I have fine tuned my current electronic transformer (the one where I have wound myself two secondary windings of 4 turns each). Based on my calculation, output was between 9.5 and 9.8 (worst-best case scenario). After removing 2 turns from the primary (easy!!!), from 67 to 65, I should have 9.8 to 10.1, which I consider as perfection when working with so many unknown variables and few secondary turns.

No problem in operation, reading on DMM is not logical (increased from 15.8 to 17V), but the light is slightly less subdued and it seems just right. I can still look at the filament and it is bright yellow. It seems that both tubes have increased emission, since the cathode voltage is slightly higher (but that is something I should measure within the shortest period of time, not "tomorrow". Anyway, the sound seems to get that extra edge of precision and speed.

Edit: strange how I started wrongly using taught instead of thought as past form of to think?! I saw it in a couple of forums used by native English speakers and started using it without too much of a thought. :( Still, something did not seem right, and I had to check... thus here we are, editing.

#107

Posted: Wed Jun 18, 2014 2:50 pm
by Paul Barker
some native English speakers speak the th sound as a t. But not the Queen.

#108

Posted: Wed Jun 18, 2014 3:48 pm
by shane
That's the effect this forum has on people, Alex. Your native tongue may be Serbian, but you'll end up speaking it with a Yorkshire accent!

#109

Posted: Sat Jun 21, 2014 10:50 am
by Alex Kitic
I am considering adding a 3rd secondary for the driver tubes, most probably 3 turns for 7.5V and drop the voltage via 1.2 ohm resistor approximately for 6.3V ...

Since the driver tubes are indirectly heated, there should be no adverse effects, and the 7W power is not significant when compared to the 100W power already used for the two output tubes (the unit is for 160W).

I was also considering making 5 turns instead for 12.5V and rewiring the driver tubes for 12.6V operation, but that would imply using only 12SN7 instead of 6SN7 as the cathode follower tube... I will most probably stay with 3 turns, but the 5 turns alternative means 1/2 the current draw.

Image

The simplicity of it... a small box, a cap... and nothing to worry about (?)

#110

Posted: Sat Jun 21, 2014 11:00 am
by Paul Barker
Yes makes sense.

I bought a pair of cheap 105 watt ones £6 each delivered.

They are like the cheap local ones you found. The output transformer is large enough to work with. Like with yours there seems to be a simpler component count.

It would be a lot easier to adapt, you probably could change the frequency on them.

I realise that with a pi filter incorperating a regulator on the input voltage to reduce it the output could easily be adapted to 10v and the power rating means it would heat an 833a. This is now really helpful as it is a massive step in miniaturisation for an 833a amplifier. Rediculously small and light filament transformer.

Obviously it remains to be seen how it sounds.

I might get round to making my 833a amplifier into monoblocks with one of these, because my present switched mode heater supply is large and heavy.

#111

Posted: Sat Jun 21, 2014 11:26 am
by Alex Kitic
Paul Barker wrote: It would be a lot easier to adapt, you probably could change the frequency on them.
Changing the frequency is not too simple, but not too complicated either. Still, the black one in the picture seems to be operating at 30kHz, and I cannot hear any difference between it and the white Vossloch-Schwabe units which are more complicated, sophisticated, and operate at higher frequency (probably 48-60kHz).

My theory is that if 30kHz is 1st harmonic, the residual 2nd harmonic is as high as 60kHz, which is too high for the output transformer anyway, if you ask me...
Paul Barker wrote: I realise that with a pi filter incorperating a regulator on the input voltage to reduce it the output could easily be adapted to 10v and the power rating means it would heat an 833a. This is now really helpful as it is a massive step in miniaturisation for an 833a amplifier. Rediculously small and light filament transformer.
It is not exactly that you will need a pi filter for the 833 - the cap will do enough for the 100Hz ripple, making it inaudible. Output reduction should be performed in that case by changing the number of secondary turns, and eventually fine-tuning by reduction of primary turns numbers.
Paul Barker wrote: Obviously it remains to be seen how it sounds.
If you ask me, it sounds great - it sounded slightly cleaner and faster than standard AC from the beginning, and now that the 200Hz buzz (2nd harmonic of 100Hz) is gone, I am completely taken with this principle for tubes heating.

But I am looking forward to hearing your impressions with your several tubes prototype - while you can take my experience for granting in cases where large tubes like the 813, 833, or GM70/GK71 are to be heated.

Paul Barker wrote: I might get round to making my 833a amplifier into monoblocks with one of these, because my present switched mode heater supply is large and heavy.
Should be easy to do, and a fast way to try it... I a sure it is a winner compared to the SMPS - if nothing else, it is simpler, and smaller, and even more efficient (and durable? perhaps, since it is simpler).

I would like to suggest a durability oriented tweak: if possible, heatsinks on the output transistors (care should be taken to isolate them): they don't need much, and it is obvious that this is the only component (aside from the diode bridge) that is under stress in this application.

As for the diode bridge, I guess an NTC in series with the mains input would be in order (I have yet to install one on mine, but it's "scheduled"). I guess that even the size is not critical, and should be estimated based on the current capability of the diodes, and the current surge caused by the input capacitor (could be up to 100A - but you need to enable just 400mA constant current for standard operation at 100W after the initial surge).

#112

Posted: Sat Jun 21, 2014 11:31 am
by Paul Barker
Yes I get all that.

Agree when using near max dissipation worth heat protecting the transistors.

#113

Posted: Mon Jun 23, 2014 11:47 am
by Alex Kitic
I have modified further the electronic transformer I am currently using with the RH813, adding a 3rd winding to heat the drivers at 6.3V

Here's a nice shot of the unit before the last mod:
Image

Calculation - assuming 2.5V approx. per turn (used 4 turns for 10V), 3 turns equal 7.5V; drivers current draw 0.9A for 6.3V; 7.5V-6.3V=1.2V/0.9=1.333ohm.

Shown here the 6201 heated by what is calculated to be 6.3V:
Image

For comparison, 6201 in my preamp, heated by 12.24V DC (i.e. 6.12V per single triode):
Image

Looks almost identical to me. The pictures were taken at the same time, one after the other.

As for the electronic transformer, it does not seem to be heating more than previously, there are no audible artifacts, etc. The sound is simply great as usual... if I had a sticker for "OK, QC Pass" I would have to stick it on the unit.

#114

Posted: Mon Jun 23, 2014 12:55 pm
by Paul Barker
Great.