Cloning an HP E3615A

[just_fib_it]

Having successfully blinked an LED, I'm now about to embark on my first serious electronics project: building/cloning an HP E361xA power supply and hopefully extending it, once I've figured out how and why it all works.

I'm looking at the E361xA (x being 4..7) range of power supplies, which are effectively beefed up versions of the venerable E3610A and its higher voltage siblings. The service manual (with full schematics) is available here: http://www.physics.fsu.edu/users/Wahl/labmanuals/instruments/ps/AgilentE361516AManual.pdf

I'm currently shopping for the right components and am having some problems finding the exact values for the current shunts and the current balancing resistors for the series-pass transistor pair.

The original design uses the same resistance values for the current shunt (R58,R59) and the current balancing resistors (R1,R2). Is there a good reason for that? It doesn't seem to be about reducing the number of different components on the BOM, since the power ratings are different (10W vs. 5W).

The voltage drop over the current shunt (R58,R59) is scaled to 0..5V by R74,R75, so I assume I need to adjust those if I change R58,R59. Are there any other components affected by changing them?

If I change the value of R1,R2, do I need to scale any other component values accordingly? As far as I can tell the answer is no?

[megajocke]

Sounds like an interesting project. Using slightly different values should work, and as far as I can tell I think you are right that changing R74-75 to match should be enough.

That they are scaling all the resistor values roughly proportionally to output voltage makes sense for a few reasons. Thermal stability of the current sharing and current offset error as a percentage of rated current are some of the parameters that are kept roughly equal then for all models.

Note that the current shunt resistors are run well below their rated power dissipation. This keeps the temperature rise of the current shunt low which minimizes drift due to resistance temperature coefficient in the constant current mode.

[poorchava]

That could be quite possible if you manage to get an appropriate mains transformer custom-wound for you. Even with a single primary for a chosem voltage, you still need 5 secondaries. I don't know where does the OP live, but here in Poland there are workshops that accept single orders for custom toroidal transformers (for example audio).

Few years back I've ordered one that was 300VA and had 8 secondaries and paid around PLN 140 with postage (~35€ / $45 ). If that's an attractive price for you I could help you get one.

[Harvs]

That could be quite possible if you manage to get an appropriate mains transformer custom-wound for you. Even with a single primary for a chosem voltage, you still need 5 secondaries. I don't know where does the OP live, but here in Poland there are workshops that accept single orders for custom toroidal transformers (for example audio).

Few years back I've ordered one that was 300VA and had 8 secondaries and paid around PLN 140 with postage (~35€ / $45 ). If that's an attractive price for you I could help you get one.

Or just wind it yourself if you can't find someone to do it.  Get a torroid of the correct rating, and preferably with at least one of the secondary voltages you need, then just pull off the outer insulation layer and wind on enameled wire until you get the secondary voltages you need.  There's calculators online to get the wire sizes you need for current.  As you're not touching the primary to secondary insulation there's no issue with isolation.

[JJalling]

Having successfully blinked an LED, I'm now about to embark on my first serious electronics project: building/cloning an HP E361xA power supply and hopefully extending it, once I've figured out how and why it all works.

I'm looking at the E361xA (x being 4..7) range of power supplies, which are effectively beefed up versions of the venerable E3610A and its higher voltage siblings. The service manual (with full schematics) is available here: http://www.physics.fsu.edu/users/Wahl/labmanuals/instruments/ps/AgilentE361516AManual.pdf

I'm currently shopping for the right components and am having some problems finding the exact values for the current shunts and the current balancing resistors for the series-pass transistor pair.

The original design uses the same resistance values for the current shunt (R58,R59) and the current balancing resistors (R1,R2). Is there a good reason for that? It doesn't seem to be about reducing the number of different components on the BOM, since the power ratings are different (10W vs. 5W).

The voltage drop over the current shunt (R58,R59) is scaled to 0..5V by R74,R75, so I assume I need to adjust those if I change R58,R59. Are there any other components affected by changing them?

If I change the value of R1,R2, do I need to scale any other component values accordingly? As far as I can tell the answer is no?

How about taking a look at Gerry Sweeneys PSU project at http://gerrysweeney.com/ ? I think it has great potential.

BR Jonas

[just_fib_it]

@megajocke: In the end I found some resistors that are pretty decent matches. Most problematic were those for the E3616A: instead of 0.6R I ended up with 0.604 (5W) and 0.619 (10W). I guess that shouldn't cause any problems 

@poorchava: Thanks for the offer  I actually moved to Hungary a while ago, and I think I should be able to get some transformers custom made here for prices comparable to those in Poland.

For initial development though I'm planning to use a transformer I found on a local auction site here. The HP service manuals unfortunately don't contain any info about the transformer. They use three windings for the main circuit and use some SCRs to enable one or both of the additional taps, and have two more secondaries for the various bias and reference voltages.
I think the transformer I found is a pretty decent match for an E3615A (20V/3A) or E3616A (35V/1.7A). It has four 16VAC 4A secondaries plus a bunch of auxiliary ones. My only worry is that perhaps the winding for the +15/-12V circuit is a bit weak. It's only rated for 300mA. I must confess that I have absolutely no clue how much current an opamp would need. What do you guys think, is that 300mA enough to power both the +15V and -12V rails?


@Jonas: I think I've seen all of Gerry's PSU videos, the repair videos as well as those about his own PSU project, some really interesting stuff there!
I'm afraid I'm not quite ready yet to build my own PSU from the ground up, like Gerry is doing. That's why I figured I'd start with copying an existing design and make some gradual changes once (if...) I get it all to work.

Some of the changes I've been thinking about are:

• digital programmability using an ADC and DAC (now there's a novel idea  )
• updating some components to more modern (and more obtainable) versions. In particular the OVP circuit, which currently uses an overvoltage sensing IC that is rapidly turning into unobtanium...
• perhaps turning the whole thing into an E3631A-ish configuration with three independent output, basically combining an 8V/6A model with two 35V/1.7A models.

Thanks for all the help so far! I've ordered a whole bunch of components from Mouser, can't wait for them to arrive and actually start building something!

[sorin]

have a look  at the E3632A
Use mosfets, I think better performance

[just_fib_it]

have a look  at the E3632A
Use mosfets, I think better performance

The E361xA designs I'm looking at use MOSFETs too; the low voltage models (E3614A and 15) use two IRFP044s, the higher voltage models 16 and 17 use IRF440s (at least I think so, the service manual only contains an HP part number  ).

[megajocke]

For initial development though I'm planning to use a transformer I found on a local auction site here. The HP service manuals unfortunately don't contain any info about the transformer. They use three windings for the main circuit and use some SCRs to enable one or both of the additional taps, and have two more secondaries for the various bias and reference voltages.
I think the transformer I found is a pretty decent match for an E3615A (20V/3A) or E3616A (35V/1.7A). It has four 16VAC 4A secondaries plus a bunch of auxiliary ones. My only worry is that perhaps the winding for the +15/-12V circuit is a bit weak. It's only rated for 300mA. I must confess that I have absolutely no clue how much current an opamp would need. What do you guys think, is that 300mA enough to power both the +15V and -12V rails?

A 300 mA winding should be enough for the control circuitry. How much voltage does it have? The control circuitry doesn't need more than a few tens of milliamps from each of the positive and negative rails.

With four 16V 4A secondaries (assuming rated at nominal line and full resistive load) the transformer should be suitable for around 70V DC 2A maximum output at low mains (-10%). At low current and/or nominal or higher mains you might be able to get more voltage.

Figure A-2 in the manual shows how the taps are switched. For the N1 (middle) and N2 sections you'd use a single 16V secondary for each. For the N3 section you'd use two 16V secondaries in series. Then the tap selector can choose 16, 32, 48 or 64V AC RMS for rectification.

Such a configuration would need more heatsinking than the original power supply however, but it should still be well within the capabilities of a pair of typical TO-3 (or TO-247) mosfets as is used in the original as pass transistors. About 50 W total needs to be dissipated at high line and full output current when the output is shorted or just above a switching point.

[just_fib_it]

A 300 mA winding should be enough for the control circuitry. How much voltage does it have? The control circuitry doesn't need more than a few tens of milliamps from each of the positive and negative rails.

There's actually two of them: 25V and 43V AC, both rated for 300mA. I was already considering using both of them with a center tap, but if the control circuitry only needs that little power I guess I can simply stay with the HP design and only use one winding. The transformer I got also has an 8V 1.5A winding which should be a good match for the 5V circuit.

Speaking of (pulse) transformers: I've been looking at the SCR-based pregulator design and can't seem to figure out why HP use two different mechanisms to drive the SCRs. The high voltage variants use an opti-isolated triac driver IC (MCP3020Z http://www.alldatasheet.com/datasheet-pdf/pdf/163953/ETC/MCP3020.html) whereas the low voltage ones use pulse transformers (the service manual doesn't go into a lot of detail on them, only that their primary inductance is 5mH.

Both versions use the same SCRs and the same circuitry to control the SCR drivers, only the part in between ("SCR firing circuit" on the schematic linked in the first post) is different. Do SCRs need different firing pulses depending on the voltage/current going through them?

[megajocke]

Hmm, maybe it's because the opto-triac version needs a few volts over the thyristor to be able to trigger the gate. But as soon as the thyrisor triggers the voltage drop of both circuits will be the same. This might give the pulse-transformer version a slight advantage in the low voltage versions, but I'm not sure how much without trying or simulating it.

You could use your 43 and 25 V windings in series but out of phase to give something like 18 V AC. The 7915 is only rated for 35 V maximum input so it would be unhappy if used with the 25 V winding which could give upwards of 40 V rectified at high mains and low load.

[poorchava]

I think you'll be perfectly fine having 300ma for bias and control circuitry.  Opamps typically consume up to a few miliamps plus load,  so as long as you keep the link ad low you should be fine.