I am thinking of having PCB's made for this regulator using the Supertex LR8N3 -- size will be about 3" x 2 " -- wondering whether, in fact, 330u/450V isn't overkill. Bypass caps are WIMA MKP's (but choice is up to you.) I estimate the cost (net of shipping) to be something less than $3.00/board. Note that on the PCB I alot space for the MJE3439 or a BU205A. Email me if interested: jack@tech-diy.com I haven't sent the design out so am awaiting suggestions. As usual, but not always, this is a Mother Theresa project.
ps -- about 1/2 of the Last-PAS regulator boards are gone.
ps -- about 1/2 of the Last-PAS regulator boards are gone.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
3D Rendering:
An externally hosted image should be here but it was not working when we last tested it.
Do you really need a double sided board?
Why don't you draw wide polygons in order to use as much copper area as possible and reduce the amount of wasted copper (particularly useful for DIY PCB manufacturing)?
Do you really need two transistor mounting locations? I would just put the TO-220 or TO-126 and the TO-247 or TO-263 one over the other and retouch the pads a bit...
Why don't you draw wide polygons in order to use as much copper area as possible and reduce the amount of wasted copper (particularly useful for DIY PCB manufacturing)?
Do you really need two transistor mounting locations? I would just put the TO-220 or TO-126 and the TO-247 or TO-263 one over the other and retouch the pads a bit...
EVA:
I hadn't drawn in the powerplane -- usually I set this as a different NET than "0" and connect at one locus. I have found these designs to be very, very quiet and stable using this grounding methodology.
The TO-247 case for the BU205 (actually TO-218) is a different pinout than the MJE3439. The MJE3439 is ECB, the BU205 is BCE. I guess you could "flop" one or the other -- but they must be heatsinked in any case.
btw, the LR8N3 costs about $0.65
I hadn't drawn in the powerplane -- usually I set this as a different NET than "0" and connect at one locus. I have found these designs to be very, very quiet and stable using this grounding methodology.
The TO-247 case for the BU205 (actually TO-218) is a different pinout than the MJE3439. The MJE3439 is ECB, the BU205 is BCE. I guess you could "flop" one or the other -- but they must be heatsinked in any case.
btw, the LR8N3 costs about $0.65
I think it's OK, but only one advice:
The ground pin on R2 is getting awfully close to the 280V regulated output trace. I advice a little more spacing there... 3 millimeters at least to be on the safe side.
Second, but this is for your personal taste only: This layout can be done on a single layer... Do save your manufacturing costs and try to work those two traces to the soldering side....
Cheers
The ground pin on R2 is getting awfully close to the 280V regulated output trace. I advice a little more spacing there... 3 millimeters at least to be on the safe side.
Second, but this is for your personal taste only: This layout can be done on a single layer... Do save your manufacturing costs and try to work those two traces to the soldering side....
Cheers
ok, flop the MJE3439 -- it takes a smallish heat sink anyway -- the PCB is about 0.25 inch wider (adds about $0.07 to the cost):
An externally hosted image should be here but it was not working when we last tested it.
that looks just fine... got no more comments 🙂
there is no difference in price between double sided pcb's or single sided??
there is no difference in price between double sided pcb's or single sided??
how "HV"? - Is there any creepage spacing consideration
given some trace separation requirement it probably helps to use opposite sides of the board for the 2 unreg input traces and large ground plane clearances everywhere
given some trace separation requirement it probably helps to use opposite sides of the board for the 2 unreg input traces and large ground plane clearances everywhere
here's what the first protoype looks like == there are some things I don't like -- the connectors are too close to the mounting holes (even though I bought some nylon spacers, it wouldn't be my default choice). I used 1.0 uF 400V WIMA PP on the output == there is about 50mV of "noise" (on 292 VDC out) == but it does take the wrinkles out. FWIW the "Last PAS" regulator has much less noise (but is much, much more expensive):
An externally hosted image should be here but it was not working when we last tested it.
that should help the lower the hf impedance -- since R1 is a pretty large value, I am thinking to place an electrolytic across it (and protection diode):
An externally hosted image should be here but it was not working when we last tested it.
C3 will provide a more stable regulator, but 100uF is maybe a bit overkill since there is no load there. 10uF Low Impedance cap will be sufficient there i guess
I assume R3 is there to discharge the lot when power is removed?
For the rest, I hink this is looking quite nice...
I assume R3 is there to discharge the lot when power is removed?
For the rest, I hink this is looking quite nice...
C3 is reducing the noise gain of the regulator from 250 down to unity, no wonder the original circuit had such high output noise.
A couple of minor points:
How about a bleeder resistor for that cap?
Reverse diodes across regulators have been known to rectify RF.
With that much front end capacitance, you may want to add an inrush limiter. Easily accomplished with a power MOSFET, 15V zener and, a couple of resistors.
As for the comment about double sided boards, most quick turn board houses start with a double sided board process anyway. There is little, if any, real cost advantage.
Other than those minor points, very fine looking project.
How about a bleeder resistor for that cap?
Reverse diodes across regulators have been known to rectify RF.
With that much front end capacitance, you may want to add an inrush limiter. Easily accomplished with a power MOSFET, 15V zener and, a couple of resistors.
As for the comment about double sided boards, most quick turn board houses start with a double sided board process anyway. There is little, if any, real cost advantage.
Other than those minor points, very fine looking project.
- Status
- Not open for further replies.