DHT Heater using VCCS - PCB or Kit (Possible Group buy)

[Mike H]

Each channel has four 300B valves.

Blimey!

[Andrew]

I'd like to share my working on transformer selection, in the hope any errors can get pointed out before I get too far. Taking as example the 45 valve pair.

Vf = 2.5v
If = 1.5A
Required V supply to DC board = 2.5v + 4v = 6.5v
TX rms required = 6.5v / 1.41 = 4.6vrms minimum

Hammond 162H10 (PCB Mount type) has 2 secondary windings each 5v rated at 2A (TX VA rating is 20 VA total).
The DC Board will dissipate (5*1.41 - 2.5) * 1.5 = 6.8W

Using 63mm heatsink with thermal resistivity of 4.5 degC/W and adding 1 for mica tab and paste. Not sure if the mica tab is required with the small boards ?

Junction Temperature = Ambient + DISS * Total Thermal Resistivity
Jt = 50 + 6.8 * 5.5 = 88 degC

The heatsink bit is complicated because as I understand there are 2 transistors on the same heatsink. My poor little head breaks at that point, but I figure that I just need to worry about the TOTAL dissipation and the fact that it's split over 2 transistors will help not make it worse.

A 5 volt TX sounds fine to me. Have you considered using a generic toroid from Rapid or likewise, cheaper than Hammond perhaps?

Yes, just focus on the 7.05v (in) - 2.5v (cathode) * 1.5 Amps for the heatsink. The power won't disappear in the circuit, otherwise that would be a very clever circuit, so I make that (Vin - Vcathode) * Amps therefore 6.8 Watts across the heatsink, same as you.

Andrew

[Nick]

Assuming a 6v TX (because they are off the shelf), that gives a

maximum DC of 6*1.41 = 8.5v,

voltage drop 2.5 = 8.5 = 6.5v

PDiss = 6.5 * 1.5 = 10

4.5 C/W and 50 ambiant gives a heatsink temp of 95C (I would try and achieve a lower ambiant, thats a hot box you have there).

Assuming the dissipation is equal for the two devices, 1084 has a chip to case resistance of 0.65C/W (didn’t bother working out the control dissipation) and assume 1C/W for the mounting, that gives a chip temp of 95 + 1.65 * 5 = 104C

Hot but inside all the numbers for the device, bringing ambient down would improve that a lot.

[Andrew]

If you used a 6v TX it would be pretty equal across both devices A 5v tx, if you can get one, will generate less heat.

I agree with Nick, 50 deg C is 25 deg above summer room temps, ok, slightly less than 25 in hotter climates, but still, that's a hot box.

Andrew

[Ray P]

Each channel has four 300B valves.

Blimey!

Why shocked Mike, you saw and heard them at the last Owston meet...

[Mike H]

Yes you're right I did!

It was the financial aspect I was thinking about.

[Ray P]

It was the financial aspect I was thinking about.

I can think of more expensive vices...

No Dave, not that sort of vice.

[RhythMick]

50 degrees ambient was plucked out of the air. Nothing measured. I was trying to consider a worst case.

[Nick]

50 degrees ambient was plucked out of the air. Nothing measured. I was trying to consider a worst case.

All good then.

[RhythMick]

In my SE amps I have a pair of EML 45 Solid Plate. The box contains a warning against using CCS for the heaters.

The EML site contains this : http://www.emissionlabs.com/Articles/TE ... ources.htm, which basically repeats the warning.

Andrew's board is though a Voltage Controlled Current Source - so will self-adjust to supply the set voltage. I think that takes away some of the EML concerns - though I have no doubt they would consider the warranty void (I wonder how many warranties are ever supported, but that's a different question). Interested to hear views.

They also reference a diagram by Jac showing good and bad ways to connect DC sources:

I'm not sure I get the "good" diagram. By definition using DC heaters the + side of the filament is more positive than the - side, and I assume will conduct more, and yet they've shown a pair of identical resistors creating a virtual centre tap to ground ?

Do you guys take steps to reverse the +/- connections occasionally to even out the usage of the filament?

[RhythMick]

and this from the EML site shows the higher current at the "hot" end ... (example is 300B)

[Nick]

Hmm, yes, that EML doc IMHO starts talking sense and gets less sensible a it goes on. The point about current sources is the reason my Andrew uses the circuit he does, unlike other current sources out there it is voltage controlled, so will avoid the problem.

It then goes to talk about the difference between voltage and current indirect heated valves. MJ did a lot of work in his book investigating this and could find no difference in the I/V behaviour of the two sorts so concluded that there was no actual constructional difference between them.

And finally, the "Hot End" Well, ok that seems to be based on the assumption that the bias at one end is less than the other, but that will always be the case (if its true) with DC heaters, I notice they dont suggest any alternative. I would always use a pair of resistors to a centre tap (as Jac suggests), but that’s for improved common mode noise rejection. The valve won't know what’s outside its pins with any connection. If we look at the higher current suggestion, lets assume a worst case, only the first 10% of the cathode at the lowest voltage is passing all the current. So that’s (say) 70ma for a 300b. The heater current is 1.2A, so one end is passing 1.27A and the other 1.2A, so that’s 5.8% more current, unlikely to make any difference. But there is no way of preventing this (is if happens). I don;t think it happens, as it seems to be based on the idea that the electrons leave the cathode and go directly to the anode. They don’t, they leave the cathode and enter the space cloud, they then leave the outer surface of the space cloud to the anode. Now the space cloud is a sea of electrons, so if one end is at a higher voltage than the other, the electrons will move to equalise the voltage. Or at least that’s what I think.

[RhythMick]

Thanks Nick. As I typed it I was thinking exactly the same about the space charge. My guess is the same, that the space charge will even out any difference.

[RhythMick]

What governs the choice of resistors to create the virtual cathode. A pair of 10r would dissipate almost no power. Is there a reason to choose another value?

[Nick]

What governs the choice of resistors to create the virtual cathode. A pair of 10r would dissipate almost no power. Is there a reason to choose another value?

a couple of 10R or 22R would be my choice in most cases.