HF AC DHT heating

For the three and more legged things
Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Thu Jun 05, 2014 5:37 pm

I have just tried measuring the voltage on the 813 heaters with the simplest possible analog meter. Based on calculation and reason I expect the value to be around 10V, between 9.7 and 10.3V.

As previously mentioned, my DMM always gives a reading between 23.2 and 24V, which I imagine to be Vpp and divide by a similar coefficient as the scope interpretation of the waveform. EDIT: based on internet sources, DMMs usually mistake when measuring square wave signals, it should be factored in the T/t of the wave (probably 1.15-1.2) and the mistake at high frequency can reach 46% at 100kHz. Quickly: 24V / 1.2 = 20 x 46% = 9.2V at 100kHz, probably more (less error) at 50kHz. Inconsistent, but it boils down to around 10V, as expected. EDIT 2: this is even more stupid actually, it could be the other way round... useless, really, my first assumption looks better...

While it seems that the best method is calculation, I am still searching for some expedient and cheap way of getting a consistent reading to be interpreted... in some consistent manner.

The analog meter reads "exactly 5V" on either channel, and repets the same reading with every setting (10V, 50V, etc). The meter is otherwise rather correct with AC (mains, other filaments) but quite imprecise with DC readings (bias at 50V instead of around 60, etc.)

While the "not hesitant" 5V looks interesting (1/2 of the expected value), I wonder what would happen if I change the diode inside, replacing the 1N4007 with a Schottky (it should not be used above 45V, of course, but it might work for filaments). Opinions about the reading?

I know, many excuses to avoid buying a scope or a good Fluke (if I had the funds, I would probably stroll to some shop).

Btw, this is the meter (don't laugh):
(See attachment, EDIT)
As usual, I did not link as I wanted, but you can see the DMM if you follow the link.
Interesting enough, the site where I have found the picture states precision up to 50-60V, and it did "precisely measure various batteries...
Attachments
rolson-analogue-multimeter-yx-1000a-320-p.jpg

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Thu Jun 05, 2014 7:13 pm

Guess what?
Original output 11.5V at 60W. Removed 3 turns of 8. 11.5 : 8 × 5 = 7.18

Disconnected the cap on one unit, filament rather pale, measured as 3.6V (very slightly above 3.5). 3.6V x 2 = 7.2V

The unit with cap again measured at "exactly" 5V.

Houston, we have the filter box now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Fri Jun 06, 2014 10:59 pm

I bought one extremely cheap electronic transformer to test... 60W unit, no additional spec, 12V nominal output. It's not even bolted but sealed, not difficult to open anyway.

Good thing is that it works with a 10W automotive bulb! Extremely simple circuit, bare of all protection or filtering, except the stuff necessary for operation. The output transformer is not toroid and will be fiddly to adjust, but at the price I can play with that.

Bad thing is that the cheap and simple multimeter was not giving a logical reading. My DMM puts it at 45V (45÷2.39=18.28V) with one 10W bulb, which does not make sense.
With the diode and cap device I used before, 15.3V DC, (15.3÷1.1=13.9V AC?) ... the same diode and cap device measures 11.3V DC with my 813 heaters, for a reasonable 10.27V AC.

It seems that after all, the only way to measure electronic transformer output is either a scope, or some fancy Fluke or similar that can measure frequency up to 100kHz.

The cheap device looks actually very sound if we forget about filters and similar stuff. The transistors are grouped on one end, making it easier to heatsink. Next week I'll try to do something more with the unit, I would like to convert it to 3x 6.3V ...

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 9:58 am

OK so my hurried attempt to get a multi secondary working prior to Owston flopped.

One problem is taking too much out of the original well made product and fiddling with tiny little wires and stuff.

So I found in my stash a decent sized torroid, which I shall put on a primary that picks up where the secondary of the original left off. I shall increase the turns per volt so that I have a number divisible to the tolerance I seek for 2.5v and 6.3v, I can live with 5% on these cheap valves chosen for the experiment (1619, 6b4g, 6em7).

The use of a larger off board IT means a single unit can provide for whole stereo amp in this instance. I'll be using about 50 watts from a 60 watt capable gadget.

The biggest problems before were

1/ not enough space on the tiny output transformer to fit multiple secondaries.
2/ not enough use of the available power. These don't behave well at 20 watts.
3/ adding volt dropping power resistors adds to space issues. Alex found it not an issue to raise the dc voltage. So I have an RS order due today I have added a pair of 330uF caps to that. Might get bye with just one. I'll see if it works with just that taken off board by two fly leeds. Then I can keep the electronic transformer in it's packaging.

Press on.

Image

The circuit board in the front of the picture is just one I rob parts from. In the event because I have a new decent sized cap coming there is nothing on it relative to the project.
I can do it right or I can do it now, but I can't do it right now.

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 10:14 am

Stage one will involve intercepting the board with the cap. hooking it up to a 50 watt lamp, which admitedly will burn bright. Scope the output to establish turns per volt as it stands.

Stage two is to wind the external IT with a greater number of turns per volt so I can make a suitable division of voltage. (because in original form they are less than a turn per volt) Wind an equal secondary on the IT temporarily, fire up the lamp again so that I can establish the ratio of the losses, which will be two fold.

From that I shall have a secondary turns per volt formula, and I can wind it to power four of 1619's two 6b4gs and two 6em7's. The 6em7's shall share the 6b4g secondary in parallel.

Step by step we will arrive at success in terms of heating. Then we shall be in a position to check how it sounds.
I can do it right or I can do it now, but I can't do it right now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Tue Jun 10, 2014 12:14 pm

Paul Barker wrote:I can live with 5% on these cheap valves chosen for the experiment (1619, 6b4g, 6em7).
I would rather have more precision on any tube - and provided you can measure it, you should not have particular problems to reduce voltage applying a low value resistor in series.
Paul Barker wrote: 1/ not enough space on the tiny output transformer to fit multiple secondaries.
I find it more of a problem that all units I am able to source (at the moment, still searching) have EI ferrite transformers. In particular, the very cheap unit has a rather small transformer, and I have yet to try unwinding the secondary to see what can be done with it...
Paul Barker wrote: 2/ not enough use of the available power. These don't behave well at 20 watts.
In particular, the very cheap unit works with just one 10W automotive bulb, while adding another does not change the light intensity on the bulbs... I feel that 105W units are better quality, and so far several I found are rated for 20W minimum - like the Vossloh Schwabe units that I use in the RH813, and some Osram ET Parrot units I can buy locally.

Anyway, the units I am using to heat the 813s work even with one 10W bulb (tried)...
Paul Barker wrote: 3/ ... Alex found it not an issue to raise the dc voltage. So I have an RS order due today I have added a pair of 330uF caps to that. Might get bye with just one...
Yes, one 330uF cap per unit works just fine and besides being too large to fit inside, there are no other issues, and it is the simplest tweak to implement.

Strangely enough, the 330uF cap costs more than 1x 100uF cap, 2x 10uF cap, LM317, and 2x resistor. The purpose of the 100uF cap is to reduce the ripple to "usable" levels, followed by the LM317 set for the first lower voltage possible (probably 280V or so), one 10uF across the lower resistor to improve ripple rejection, and the last 10uF on the output. This circuit provides basically no ripple at the output (i.e. not buzz even in theory, since the buzz is a byproduct of the 100Hz ripple) and also regulates the DC, making the output of the electronic transformer regulated as well.

Because the output of the electronic transformer depends on the DC voltage and the number of turns in the first small transformer in the half-bridge circuit, and of course the type of transistors used... but since we are not going to change any of the other circuitry, regulating the DC at the input regulates the output as well. What you set will remain the output regardless of mains voltage in the area.

Of course, this additional circuitry cannot be placed in the same box, and is not really necessary. But having a regulated output voltage is probably beneficial, and besides the slight complication of soldering all these parts on some small board, the cost of the parts is probably lower than the cost of the 300uF cap which is needed for the simple tweak - while the result is much lower ripple (we pass from i.e. 7V to maybe 7mV or less).

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 12:14 pm

RS delivery contained half of what I ordered, the half without any caps in it.

So looks like I'll be fudging it with equipment pulls as per usual.
I can do it right or I can do it now, but I can't do it right now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Tue Jun 10, 2014 12:38 pm

Interesting article showing waveforms and changes in waveform (oscillation frequency) depending on load - in the documented case those are various LEDs - which are obviously much different load than tube filaments (equating basically to a halogen light filament).

Various Transformers with LEDs

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 4:02 pm

OK that is as much effort as I want to put in today. But I have a halfway house result which encourages pressing on to the goal.

The IT idea has worked very well. the voltages I predicted have come out slap bang on. With an expected output using the cap alone to smooth was 16.8v under 50 watt load. So to get a 1 tpv IT I underwound the primary to 16v and with 11 secondary turns and a 50w lamp I get 11v.

I decided against 2 tpv this time round because of the difficulty.

No issue at 2 turns my 1619's are filament starved. That is no problem at their level it gives lower distortion more gain and more impedance lower transconductance. The 6b4g and 6em7 can live with 6v quite happily.

The losses of the use of an external torroidal core is 5% losses. This is better than expected, and quite reasonable. I was worried the frequency of operation wouldn't suit the core material or the winding technique, but it seems coupling is absolutely fine.

I forgot to check the frequency. But whatever it is it'll be between 40khz and 50khz.

This is a big step forward. The electronics didn't make any mechanical noise, seems happy with the changes imposed on it.

It now remains to see if I can fit 6 secondaries for stereo ( 4 at 2 turns and 2 at 6 turns) and how they react to each other and how the whole thing sounds. But first level is to see if the heaters work. Then I have to actually complete the breadboard before I can report back on the finished result. So as I have fracture clinic and inlaws to worry about tomorrow. couple more days.
I can do it right or I can do it now, but I can't do it right now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Tue Jun 10, 2014 7:18 pm

I have found a unit with toroid output transformer. It is a cheap unit but rather seriously built (the transistors are heatsinked!). It is declared as 160W, 80W minimum. With just one 10W bulb voltage output is almost none, but with 2x20W both bulbs seem to be lit just right.

I will try to convert it to heat both 813 tubes. It should work well with 100W load. The secondary is 7 turns, several thin wires together in an insulator. I was even considering the option to unwind a few turns of primary.

What wire gauge would be adequate for 5A current in the secondary? I expect that the 10V output could be achieved with 4 or 5 turns, and if the secondary wire is not too thick I might be able to squeeze 8-10 turns.

I will probably try to adjust output voltage by regulating the DC, although I would prefer the simplicity of the one-cap solution.

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 8:12 pm

Look at it this way, the existing secondary is rated at 160 watts 12 volts which means 13 amps. So whatever they used you could use.
I can do it right or I can do it now, but I can't do it right now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Tue Jun 10, 2014 9:42 pm

I am used to secondary windings made with solid wire, but so far most HF secondary windings I saw were made with several wires, or even multiple strands. Since I am planning to use just 2x 5A, I estimate 14AWG thickness to be adequate, and I do have some multiple stranded wire with insulation that I intend to use.

As for the primary, undoing some turns might increase the V per turns ratio, but at just 100W instead of 160W I guess I can do it without particular problems.

Assuming 11.5V original, and up to 1.49x increase with the cap, I am probably going to get 16.8V just like you did. At 7 turns for 16.8V I am getting more or less 2.4V per turn. 4 turns will be 9.6, and 5 turns are 12V... thus either the caps proves to be on the higher side of efficiency, or I will have to reduce the output voltage by changing DC, or primary turns.

First the cap, then an estimate of the result, and than the modification of the secondary.
Maybe the best solution would be employing the LM317 circuitry to have a regulated and limited increase of DC, and wind 5 turns for 10V. Assuming 310V Dc for 12V output, I need to decrease the DC by 20%, using a smaller cap (100uF, or 47uF) and regulating the voltage at below the cap level with as little voltage drop as possible, to keep the regulator safe and cool?!

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Tue Jun 10, 2014 10:47 pm

My 60 watt ones are wound with two 24 AWG (not SWG) which means in this application one strand of 24 can handle 5 amps.

The reason for this is that there is very little length of wire in this application, whereas the normal convensions apply to a worst case scenario length which would mean in our case the dc resistance and so volt drop is much less than average.

two 24AWG bifilar wound is what they use for 10 amps.

Regarding Litz wire, yes it is used for HF, but this application seems not to be critical, maybe this is why they are bifilar wound?
I can do it right or I can do it now, but I can't do it right now.

Alex Kitic
Old Hand
Posts: 369
Joined: Mon Oct 14, 2013 3:20 pm

Post by Alex Kitic » Wed Jun 11, 2014 5:34 am

Paul Barker wrote:My 60 watt ones are wound with two 24 AWG (not SWG) which means in this application one strand of 24 can handle 5 amps.

two 24AWG bifilar wound is what they use for 10 amps.

Regarding Litz wire, yes it is used for HF, but this application seems not to be critical, maybe this is why they are bifilar wound?
I was actually planning to use multi-strand wire (Litz?), of a total thickness equal to 14AWG. It is definitely going to be enough, and it is rather flexible, so I should be able to wind it neatly. If it were solid 14 AWG wire, it would not be good, since the highest frequency it can pass is 6700 Hz!

The 24 AWG wire can pass frequency up to 68kHz. Since your unit was operating at 40-50kHz, the maximum wire thickness would be 23 AWG!
If you check again the pictures of the output transformer of the units I am using for the 813 currently, you shall see that it is rather thick Litz wire (able to pass very high frequency), and likewise the primary is rather thin wire as well.

While the oscillation frequency is determined in the first transformer, inadequate wire thickness in the output transformer can probably jeopardize the final result. Maybe that is what happened with your first attempt at multiple secondary?

User avatar
Paul Barker
No idea why I do this anymore
Posts: 6874
Joined: Mon May 21, 2007 9:42 pm

Post by Paul Barker » Wed Jun 11, 2014 5:46 am

Alex Kitic wrote:
The 24 AWG wire can pass frequency up to 68kHz. Since your unit was operating at 40-50kHz, the maximum wire thickness would be 23 AWG!
If you check again the pictures of the output transformer of the units I am using for the 813 currently, you shall see that it is rather thick Litz wire (able to pass very high frequency), and likewise the primary is rather thin wire as well.

While the oscillation frequency is determined in the first transformer, inadequate wire thickness in the output transformer can probably jeopardize the final result. Maybe that is what happened with your first attempt at multiple secondary?
Very interesting.

Yes maybe, but doubt it was the only factor, and I doubt it would have been a complete failure to pass any voltage, which was what happened. It didn't do anything, just shut down, despight being connected to supply.

I am not interested in picking out the nitty gritty of how one of these electronic transformers fails. It isn't my area of knowledge and not an area of knowledge I wish to spend time to aquire.

We have always found that there are mechanisms in them which frustrate our efforts at times. But once we get them working they always do work. Any which don't work we bin.
I can do it right or I can do it now, but I can't do it right now.

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest