RH2A3

If they glow, this is the place to be
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#286 Re: rh2a3

Post by Alex Kitic »

Steptoe wrote:Dear Alex, hotmail is still denying my existence but Pre65 has kindly given me your contact details and I will get the 813s in the mail now. Best wishes, Steptoe.
Thank you very much, Steptoe!
Please let me know the details - carrier and tracking number. You can post them if you wish, or maybe Pre65 can pass me the information.

One more time, thank you!
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#287

Post by Paul Barker »

Alex we do have a voltage problem. The smoothing cap on the raw dc does indeed raise the voltage considerably.

But to go to basics.

My Fluke True RMS meter makes a complete cock up of trying to measure RMS.

But my HAMEG bench voltmeter does a sterling job. The transformers I am using are rated at 11.6v with 50 watt lamp, and the HAMEG reads the RMS voltage as 11.5v. So I belive that we can rely on the readings I get from the HAMEG.

the frequency is roughly 40khz. The Fluke meter reads the frequency correctly I checked it against a frequency counter. (the one with the cool Nixie display: thanks Stuart :D ) But it isn't stable, though that may not matter.

When the cap is added the voltage goes up a great deal.

Pictures tell a thousand words.

Image

Image

Image

Image

Image

after implimentation of a 220uF smoothing cap I get

Image

At this voltage

Image

And yes the lamp is a lot brighter.

The Fluke True RMS is totally confused

Image

But it can count the frequency

Image

Image

The frequency is lower and voltage /power greater with the smoothing cap.

I'm puting my money on my HAMEG for acurate RMS power.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#288

Post by Paul Barker »

I believe the absolutely safe way to sort this out for the 813 is for me to obtain one of the transformers you have and use the HAMEG to set the voltage to 10v. Then you duplicate.

So where do I get this transformer?
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#289

Post by Paul Barker »

You will notice a discrepancy between my scope and my Hameg. I would rather put my money on the Hameg because when testing the transformer in exact normal use conditions it read a result within 1% of the stated output. The Oscilloscope isn't having a very good time with this project I am not relying on it. I think there is some thing RF going on which is confusing it. It read 18.6v RMS when the Hameg read 11.5v when the item is rated to produce 11.6v and the lamp brightness tells me of the two that the Hameg is correct.

Once the smoothing cap went in the scope showed another voltage of 16.2v but I prefer to trust the HAMEG which states 15.4v.

I am thinking the scope is crap. But I can't afford to go out buying any more scopes.

The square wave is also crap, let's be honest.

But I am sure the project is off the ground and in the air.

On my unit a reduction to 7 turns should produce 10.5v. Just on tolerance, we could probably shed the 0.5 volts in the lead. I'll try that anyway with an 813. Might be tomorrow now.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#290

Post by Alex Kitic »

Paul Barker wrote:I believe the absolutely safe way to sort this out for the 813 is for me to obtain one of the transformers you have and use the HAMEG to set the voltage to 10v. Then you duplicate.

So where do I get this transformer?
The transformer is a Vossloh-Schwabe made in Germany, Lite Line 105/12. I guess you will not find it on the local market. Besides, I think that we can work to approximate voltages with any given unit. Probably the best solution will be to work on the output transformer as well.

You actually helped a lot by pointing out that the Fluke 116 does not cut the mustard. The guy in the video uses a Fluke 87... that unit costs 4-5 times more.

I did some more thinking and thinkering and found some correlation. The voltage increase you have seen is proportionate to the reduction in ripple, i.e. the increase of DC RMS voltage at the "input" where we add the capacitor. My voltage increase is a little higher since I added a 330uF cap (I had those at hand). My increase in DC voltage is 310/208=1.49 (just measured with and without cap).

The nominal voltage was 11.5 (thus across 2x 0.15 ohm I should get 10V RMS across the heater). Voltage across the heater is slightly above 22V, and dividing by the factor 2.29 this is very near 10V (as expected). Voltage drop across the 0.15 ohm resistor shows a current draw of almost 5A (again calculated with the 2.29 factor), this sligthly lower draw corresponds to the slightly lower voltage than 10V.

On the tweaked unit voltage across the heater gets about 31V which is about 13.5 V (!) if the 2.29 factor is correct. If 13.5V, current draw is probably above 5A. Voltage drop across 0.63 ohm resistors should be at least 3.15V, meaning that the voltage at the output of the transformer should be at least 16.65 (probably more, if current is proportional it could be 6.5A, and the voltage drop up to 4.26V, meaning 17.76V!). 11.5V x 1.49 = 17.135, which we can assume as a reasonable approximation!

I would need to burn 7V, and this means more than 1 ohm resistance. Now it seems much easier to tweak the output transformer! The result I am after should be around 22-23V under load (measured with my DMM, based on the assumption that with a square wave average and RMS are the same, and since I am measuring average P-P, the correction factor 2.29 gets us the "correct reading"...

Now I need to try to tweak my output transformer... if possible (should be fairly easy if toroid).

Have you got some "normal" garden variety DMM that measures average AC values? You could check the math by comparing your HAMEG with the garden variety average unit...

I was also considering to go and buy some less sophisticated units to tweak if mine do not have easily tweakable output transformers... anyway, the video you linked to is enormously helpful (but the price of the Fluke 87 is very high).
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#291

Post by Alex Kitic »

Burning 7V at 5A current draw means dissipating 35W per channel. It's insane.

Tomorrow I will reopen the units and assess how to remove some turns. If toroid, it should be simple. Otherwise, a stroll to the nearest shop to buy something "simple".

I am also getting interested in the 2 secondary windings alternative: cutting the secondary in half and equalizing a little bit the two, with cap added, should give at least 7V each... from there to 6.3V seems an easy task: noiseless/humless 2E22, looks promising.... all from one unit.
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#292

Post by Paul Barker »

Alex Kitic wrote:Burning 7V at 5A current draw means dissipating 35W per channel. It's insane.

Tomorrow I will reopen the units and assess how to remove some turns. If toroid, it should be simple. Otherwise, a stroll to the nearest shop to buy something "simple".

I am also getting interested in the 2 secondary windings alternative: cutting the secondary in half and equalizing a little bit the two, with cap added, should give at least 7V each... from there to 6.3V seems an easy task: noiseless/humless 2E22, looks promising.... all from one unit.
You don't need to unwind just scratch off the insulation at the number of turns you want and tap off there.

On my picture
Image
it is the largest transformer, the secondary is brass coloured. It is bifilar wound, so what looks like 20 turns is just 10. Min eis glued into place so it will be a lot easier to solder in a a tap. Do it at the opposite end to the capacitor.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#293 END RESULT - HF AC

Post by Alex Kitic »

As announced, this morning I decided to go for all or nothing - either it will work, or if I ruin something - I would have to buy new ones anyway because it is neither an option to destroy tubes at extremely high filament voltage, nor to burn 35W per tube: in that case the DC voltage regulator burning about 10W per tube looks very easy. And I do not want to go back to DC, since I prefer AC (conceptually and at least for now, in terms of sound) and it looks more promising, cheap, and universally applicable.

My units (unfortunately?) look rather different in the main components. The output transformer is not toroid but rather classical ferrite. You can see the unit pictured in the attachment, the added white and blue wire lead to the cap place outside of the box (which can than be closed as if nothing happened - since I was planning on installing the two boxes on the outer side of the back of the amplifier, keeping them as cold as possible). There are additional components on the lower side of the PCB - surface mount types.
Attachments
IMG_2404_web.jpg
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#294

Post by Alex Kitic »

It is rather easy to remove the output transformer, since it is just soldered on the PCB with 4 holes (2 primary, 2 secondary terminals).

The difficulty lies in the fact that it is not a toroid. This would be rather difficult to transform into a two secondaries unit with any precision, but maybe I am wrong there. Anyway, I would prefer to operate on a toroid :D

It seems that there are 2 layers of 4 turns each, in total 8 turns. My estimate is that output voltage is about 17V, thus I need to keep 8:1.7 turns, i.e. 4.7 turns. Obviously, this means keeping 5 turns and unwinding 3.

The first is difficult to pull out, but after that it becomes much easier. Of course, this would be a great bobbin to wind two secondaries on, but the ferrite core is glued and I did not need risk braking it. If there was a way to open the ferrite core, it would be easy to to anything with the bobbin, obviously.

Thus I left 5 windings - on the picture you can see the upper 4, only the leftmost remains. My estimate is that I should get about 10.6V and I would probably have to burn some of it, but that is negligible.

After the operation, it is easy to solder back the transformer in place, and at first glance one would never know that it was tampered with (who cares).
Attachments
IMG_2410_web.jpg
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#295

Post by Paul Barker »

you could take the output at the voltage it is to a self made torroid wound with a single tapped coil, full length to the output of your gadget and tap off at the appropriate voltage. If you bought a core the same dimensions as mine you could emulate the number of turns on my secondary as your full length, then tap off where necessary.

Or buy the units I have which I linked to on eBay. OR shall I set this one up for the 813 here and then send it to you? then you can evaluate it's sound against what you have. To be sure I will use a variac to set my mains to your voltage, as my mains is 250v which may affect results.

Once we have a system on the go for the one, and it evaluates to quality of sound, I can send you another for the other channel.

These units are exactly the right size for the job. This would also give you a chance to check longevity, as you could leave it running for days.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#296

Post by Paul Barker »

As an asside, regarding using these to generate 230v AC on the output which could then be used to supply HT the obvious way is to buy a very small torroid 15v secondary, attach secondary to the output of this electronic transformer and take power out from the torroid's primary. Most torroids would be efficient at 40khz and the core size requirment becomes much less at 40khz so the small versions would be perfectly adequate.

An idea for another time.

Basically when you do your AC at HF you save on iron and copper.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8879
Joined: Mon May 21, 2007 9:42 pm

#297

Post by Paul Barker »

We posted symulatneously.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#298

Post by Alex Kitic »

OK, now the results and some standardization, if possible.

Without the cap, the light of the filaments is very poor. I forgot what I have measured with my DMM, but I guess it was somewhere around 16V... not sure, but if it matters, I can always unplug the cap and check.

With the cap, and without any resistors, the light of the filaments seems just right, comparing with the unmodified unit it was slightly less bright, athough it is difficult to assess filament voltage on TT by change in brightness (I am not a robot, after all...). The resulting voltage is around 25V -- factored with 2.29 as difference between perfect square wave and this imperfect wave, the result is 10.91 (tops). It is important to add that the 25V actually is not a fixed value I measure every time. After some heating, it falls down to 24.2 - but what is most interesting is that the difference between the two units keeps being 0.5V - I have compared the length of the wire extracted, and it is literally identical. This difference could be down to the difference in tube construction (one is RCA the other is probably a Haltron). After the bad experience of loosing one tube to filament failure, I am not so happy to remove those tubes from the sockets...

24V would equate to 10.48V - which I guess could be considered good enough to proceed without any additional resistors burning voltage.

Measured through my "diode plus cap adapter" which I have now recalculated at approximately x1.1 (1.0979 to be precise), the voltage across the heater is much more stable and without any resistors it results as 11.16V (i.e. does not change with the passage of time and the heating of the filament). With my formula, this equates to 10.14V, which is very good I guess, and the fact that it remains stable give more weight to this measurement.

With a 0.11 resistor in series (2x 0.22 in parallel) the voltage drop should be 0.55V (at 5A). I measure approximately 22.3V AC across the filament (but that is also subject to slight variation) and results in 9.73V AC (RMS calculated). The DC reading is again very stably 11.06V, equating to 10.05V (spot on!).

Whichever approximation is better, I think that probably both solutions are acceptable, but for the time being I decided to test with the 0.11 ohm resistor in series - to hear whether this rather drastic decrease in filament voltage imparts any important change to the sonic result.

What bothers me is the fact that there should be a difference of 0.55V across the resistor, and the DC measurement method shows only 0.1V... the AC (average) measurement shows a decrease about 2.5V roughly, which equates to a drop of 0.545 volts!

Thus, if I had to choose which reading to believe, it would be the AC average factored for the waveform shape (divided by 2.29 instead of 2). The ideal thing would be to have a Fluke 87 at hand... but that will have to wait. Maybe the ideal resistance to add would be 2x 0.1 ohm in parallel... or maybe it could go without any resistance. I shall keep checking the AC reading from time to time, and I am also interested in checking the temperature of the units - they were running much warmer since the addition of the caps, and I expect they will run cooler now since the current draw and general power consumption must have decreased.

As for the sound, the first impression is of a slight relaxation. It does not feel underpowered as compared to running the filaments at excessive voltage, but it does sound slightly mellower. Of course, this is just a first impression. There is probably no important change, aside from a probably much longer tube life.

The only different reading worth mentioning is a difference in bias on the RCA tube (No. 4). It used to have at least 5V more bias than the Haltron (No.1) - but now it is actually biasing at 4V less than the Haltron! Anode voltage, drop across the voltage dropping resistor between cathode and LM317 - everything is identical, but for a lower drop across the primary. This is probably due a difference in g2 current draw (more g2 current) and shows how different heater voltage must have some influence on the operation of the tube.
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#299

Post by Alex Kitic »

Paul Barker wrote: Or buy the units I have which I linked to on eBay. OR shall I set this one up for the 813 here and then send it to you? then you can evaluate it's sound against what you have. To be sure I will use a variac to set my mains to your voltage, as my mains is 250v which may affect results.

Once we have a system on the go for the one, and it evaluates to quality of sound, I can send you another for the other channel.
Thank you, Paul - while it would be interesting to check the difference in sound between two electronic transformers, I am pretty confident that it is probably not so apparent.

Of course, you have already helped more than I can say by sending tubes and transformers... and supporting my HF AC idea :)
Paul Barker wrote: These units are exactly the right size for the job. This would also give you a chance to check longevity, as you could leave it running for days.
I would rather not leave the RH813 running for days ... tube consumption aside, the heat of the tubes alone is impressive. It would probably burn my equipment rack after a while (I would have to extract it from the rack, since it is actually heating the board above it quite a lot). Heat is definitely a consideration with an amp of this type and size - although I guess mine is actually already rather small in comparison to other 813 amps (any? difficult to find on the net: but most seem to build them as mono-blocks while I have opted for the stereo solution, and mine will probably end up being the size of someone's one channel amp).
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

#300

Post by Alex Kitic »

Alex Kitic wrote: The only different reading worth mentioning is a difference in bias on the RCA tube (No. 4). It used to have at least 5V more bias than the Haltron (No.1) - but now it is actually biasing at 4V less than the Haltron! Anode voltage, drop across the voltage dropping resistor between cathode and LM317 - everything is identical, but for a lower drop across the primary. This is probably due a difference in g2 current draw (more g2 current) and shows how different heater voltage must have some influence on the operation of the tube.
I have just removed the 0.11 resistors. The voltage (AC reading) has dropped a little bit, showing just almost 21V, which is slightly above 9.1V.

Without the resistors, the heater voltage is almost identical for each valve, at approximately 23.8-23.9V, and according to my calculation this is below 10.5V.

Interestingly enough, as soon as heater voltage has increased a little bit, the bias on the RCA tube has recovered, and is now 1V above the Haltron. Total current draw and anode voltage remain identical.

As for sound and power, I have just played some Tsuyoshi Yamamoto on the edge of clipping the piano outbursts (very loud). After removing the caps, some headroom has returned, and the outbursts are less hard (less clipped, or not clipped, choose your definition).

Thus without resistors it stays until further notice. I guess some voltage is probably dropped between socket and electronic transformer terminals, although I am using relatively thick wire... it is 5A of current, after all.
Post Reply