[AndrewFischer]

Today's tubecad blog announced the Rev. B Aikido boards.

http://www.tubecad.com/2007/01/blog0096.htm

The new mono boards support a much wider range of tubes. You can also have 6 and 12 tubes on one board. The new stereo board will work as a headphone amp.

Still thinking about the 6X4 supply. 70mA * 250V is 17.5W. While that should be enough, Brooksie likes to run his tubes at high idle currents.

I'm sure there are some combinations that will work with a 6X4 and others where it won't be enough. I'll design something that will work with the 6V4 or 6CA4.

 

[AndrewFischer]

Here is the latest version of the HV supply board.

I've moved the choke off board. I'm not aware of PC mounted chokes in the 1 to 10 H range. For now, L1 is a small open frame Hammond. They are about $8 to $10. I chose some capacitors off of the UPW spec sheet, so I'd know what sizes to use on this board layout.

This needs work. I've got some silk screens overlapping. C1 is a little close to the rectifier tube. UPWs are 105C rated. I think I'll move it a little to the right and the tube a little to the left. There is wasted space at the bottom too. . .

Edit to remove old layout. New one to follow

 

[AndrewFischer]

I spent a few minutes on the layout this morning. The diode is back. I've moved the tube away from the capacitors. The board is a little smaller at 3.9"x2.7". Not done yet. The diode is cutting into a trace. Board is still too big.

 

[AndrewFischer]

I'm a EE. So far, I haven't done any circuit analysis. That just isn't right.

The HV supply will be putting out about 300V to 360V. With a 6V4 the maximum current output is 90mA. Using ohms law, the impedance of load is going to be about

300V/.090A to about 360V/.090A

or3.3K to 4K. If the current is about 35mA it will be more like 10K. anyway... The 120Hz impedance of C1 should be about 1/10 of the load or somewhere between 300Ω and 1KΩ at 120Hz.

C1 = 1/(2pi*120Hz*300Ω) to 1/(2pi*120*1000Ω) or about 4.4µF to 1.3µF

Even at the highest possible load we might see with a 6V4, 47µF is overkill.

For best results, L1 should be somewhere between 4H and 10H. As the current goes down, the choke's inductance needs to go up.

I'm going to change C1 back to 10µF. It will be easier on the tube and make little difference at the output. I may make C2 a little bigger.

 

[AndrewFischer]

I've fixed all the errors I'm aware of.

Partial bill of materials.

1. Transformer: Hammond 272BX (300V 100mA) or 273DX (350V 90mA)
2. Choke: Hammond 156L, 157M or 193B. 8H 259Ω 157M is a good fit. 193B is overkill.
3. V1: 6V4
4. C1: 10µF 450V Nichicon UPM or UPW. 22µF 450V UPW will also fit.
5. C2: 220µF 315V Nichicon UPW. 330µF 250V and 100µ 400V are alternates
6. C3: 22µF 200V UPM
7. C4: Undecided.

 

[AndrewFischer]

Updated unregulated board.

Changes:

Output bypass cap can be a Sprague 716P in the .1µF to .22µF 400VDC range or a smaller box cap.

C5, the cap on V1/4 is now a .1µF cap. Several box caps from Wima and others will fit the flexible pad sizes here. Need to fix the pour here... Oops.

Silk screen values for the resistors.
Changed diode to a 1N4007. Part is really cheap and could be left off.

Other parts shifted around to make room for the Orange Drop.


I'm still thinking about using a 6X4 rectifier. Most of the tube complements I've been looking at draw 50mA or less driving my Sennhiesers. That's well within the 70mA 6X4 limit.

But this board is more flexible. If the 90mA 6C4 isn't enough a 6CA4 drops in.

 

[AndrewFischer]

I've sent this off to Olimex for production.

Many small changes.

RN65D resistors
Holes changed to match Olimex drill rack.
Tiny changes to meet Olimex 10mil design rules.
Better silk screen text.
10mil min silk screen for Olimex.

 

[AndrewFischer]

I spent a few days looking at linear regulators suitable for an Aikido. Output between 250V-350V and 50mA-100mA.

I played SPICE with all sorts of LM317 based regulators. This is a good starting point for anyone thinking about following that route: http://www.national.com/an/LB/LB-47.pdf

In the end, I decided to go with Broskie's simple regulator. Changing the zener from 10V to 8.4 drops the current in the two resistors to 8.4mA. 15K resistors instead of 14.5K sets the output to just over 250V. The lower current reduced the heat coming from each resistor to just over 1W. The MJE350s are loafing at about 1/4W each. The TIP50 runs around 5W to maybe 9W. It will need a heat sink. The rest of the parts aren't heating up much.

I made a new layout with a decent sized heatsink for the TIP50. I left room for heat sinks on the MJEs. The resistors are bigger too.

For now, the board is single sided. Not quite ready to send out.

 

[rsabo]

Would you mind running over the calculations for the heat output of the TIP50?

Also, I don't really see the need to design a new reg board - there is one. If you think mine is inferior in whatever way, fine; however, anything else is just a duplication of effort, imo.

edit:
Quote:

The TIP50 runs around 5W to maybe 9W. It will need a heat sink.

If I'm reading the datasheet correctly - and google backs me up on this in at least one spot, in the quick time that I've checked - the TIP50 can dissipate up to 40W, meaning that there may not be a heatsink required, and if there is, a rather small one will do.

apologies for any grammatical complexity, I'm not quite awake enough yet.

 

[AndrewFischer]

Figure 4:

http://www.fairchildsemi.com/ds/TI/TIP50.pdf

The TIP 50 can dissipate 40W at 25C. As it heats up, the power drops to zero at 150C. To keep inside the operating area and have a little margin, I'd like to keep the device below about 105C.


The Fairchild data sheet doesn't give the thermal resistance of the case but the ST does.

http://www.st.com/stonline/products/...re/ds/4147.pdf Thermal data is at the top of page 2.

Thermal Resitance Junction-Ambient is 62.5°C/W. Without a heat sink my Aikido will heat the junction to about 600°C and the transistor will fry.

Thermal Resistance Junction-Case is 3.125°C/W. The thermal resistance of the heatsink is about 2.6°C/W.
http://www.aavidthermalloy.com/cgi-b...=530002b02500g

The thermal resistance of the insulator is about .67°C/W http://www.aavidthermalloy.com/produ...ilms_iso.shtml

(2.6+3.125 +.67) *10 ≈ 64°C Add in the 30 to 40 that could be inside a tube amp and the TIP50 is inside the 105C limit at 10W.

This shorter heat sink might work too: http://www.aavidthermalloy.com/cgi-b...=529802b02500g. I'd rather take up the extra space and not worry about figuring exact thermal conditions.

TIP50 Power dissipation I got from running SPICE simulations. As a rough estimate you can just take the load current. The current in the load is going through the TIP50's collector-emitter. My worst case is 100mA load, Vin = 350V and Vout = 250V. W = I*E. .100X(350-250) = 10W.

Your layout is nice and compact. The TIP50 will need an off board heat sink. It doesn't say on the data sheets, but one of the leads is likely bonded to the case. Insulate. If you stick with exact original schematic, the two 14.5K resistors need to be at least 2W devices. Also check the resistors you select are rated at full power at about 70C. If you need to make the resistors bigger, TO-220 resistors might be a way to keep board area down. Edit: with a 200V output and a reference change, you should be able to stick with 1W resistors.


I've made some small changes to the original design including a few component changes. I'm also working with a different form factor. I want these to fit in half of an Olimex SSS panel. I'm also playing with drill sizes and silk screens so I can get two boards for $27.30.

 

[AndrewFischer]

Quote:

Originally Posted by rsabo /img/forum/go_quote.gif Also, I don't really see the need to design a new reg board - there is one.

Some people watch TV or play computer games in their off time. I play SPICE and layout PC boards.

 

[AndrewFischer]

Minor changes to original design. Replace zener with IC reference. Add film cap to output. Pads to connect two mono amps. Resistor and pads for off board neon indicator. Extra gnd pad is a test point.

The board is about ready to send out. Removed redundant MJE350 heat sinks from silkscreen. Board meets Olimex rules. Size matches the raw tube supply.

 

[AndrewFischer]

Browsing the TI catalog I found the TL783. The TL783 is a high voltage 3 pin adjustable regulator. It can take a 125 Volt difference between the input and output pins. Nifty. Google found me an existing regulator design that is nearly identical to what I would have done.

http://www.gyraf.dk/gy_pd/g9/g9_sch.gif

I changed that a tiny bit and put it into Eagle.

Resistor values are for about a 250V output. The string of zeners protect the TL783 during startup. Simple. Those are 5W zeners so the regulator will still blow up if it is shorted to ground. If you build the circuit as shown, R1 and R2 don't draw enough current alone for the TL783 to operate. The output voltage will rise if the output is not connected to a load.

EDIT: Removed R3. With a tube rectifier it is not needed. The DC resistance of the choke and rectifier tube limit current surges. I did some checking and start up doesn't even get close to the 1N5365's short term power limits. 3.5A for 8.3mS .9A for 1S. I didn't check but it might even survive a brief short to ground.

 

[AndrewFischer]

3 fit on an Olimex SSS panel.

 

[AndrewFischer]

Sent to Olimex for production.




Olimex shipped the rectifier board Monday. It should be here in about a week.