My HP E3630A and me

My HP E3630A and me

A few months ago I needed to buy a new power supply to change my old self-made single output supply.

I try to find one but prices and actual quality were not satisfying me.

I am a French (sorry for my poor English!) professional electronic engineer and I thought I can find a good old one, even in a bad condition because I'll be able to repair a product with traditional components. I found a vintage 3 outputs E3630A on the secondhand market, and get it for a few money.

Good surprise! I received a well working product, in perfect cosmetic condition. Looking to the signals qualities with an oscilloscope I was very happy. With so long time from manufacturing (I think about 80's) it works fine and specifications are consistent to the technical manual.

Looking to the market again, I noticed that a similar power supply was always brand new manufactured by others firms. So it was undoubtedly a very good product, and I was really satisfied.


Before putting it on in my workbench and adopt it, I make a final look around.

First, I noticed that it should be more comfortable for trimming output voltage to have 10 turns' potentiometers, mine is only 1 turn.

Then, I remarked a small unjustified instability of the voltage display, oscillating around 2-3 points of measure. I thought it was caused by an old and bad electrolytic capacitor, but it wasn't.

Currently using it, I was surprised that as there is an auto ranging process on the voltage display (upper than 19,99V on 20V scale - in fact it supplies up to 24V regardless of 20V in manual -good news again) but not on the current (Amps) display. So, when using smalls intensity power (in fact - the most cases), while it could be displayed 0.000 to 0.500 A (on 20V scale) it only shows 00.00 to 00.50A (and 00.00 to 02.50A on 6V scale). 10mA resolution is not very precise!

Why opening the enclosure?

As I am perfectionist (nobody's perfect…), I wanted to fix this. And I suppose that others and similar power supplies can be also be modified in the same way…

The deal

Understand and fix the instability of voltage display.

Adding 1 digit of precision (1mA despite of 10mA) on the Amps display.

Making an auto-ranging Amps display despite of a fixed one to complies with 2.5A capability of 6V output.

Minimize the modification, keeping it reversible, this means keeping the originals PCBs and ICs on boards.

Replaces 2 Voltages settings 1 turn potentiometers by 10 turns' potentiometers.


Hereafter I only explain what I found and what I've done. If you want more details about the subject, please contact me freely.

I remind to everybody that I am a professional electronic engineer and that it is not (so) simple to do. Nobody will be responsible for the damages due to erratic modifications caused by your own work! If you are not friendly to electronic, don't touch anything. And always keep in mind that the supply is powered by non-protected mains power. Always disconnect the product from mains to operate!

Furthermore, I have no schematics or any documents others that the technical notice distributed by the constructor. I simply look, draw and understand how it works.


Upgrading the voltage display:

The voltmeter use 2 tensions references (200mV and 2V) and switches between them according to the tension to measure. The reason why the voltage on the display is unstable is the use of 200mV reference. The sensibility on such small value of tension added with the instability of the small tension to measure results in instability of the display.

Solution: increasing x10 the reference and x10 of the signal to measure will reduce dramatically the instability.

Then the 200mV reference is no longer used. The 2 ranges of measures are now obtain by switching a divider of the signal to measure despite of the reference voltage.

Voltage to display (V) Divided voltage range Display (Vref=2V) Auto Range
0 - 6V 0 - 600mV 0.00 - 6.00 Low
0 - 19.99V 0 - 1999mV 0.00 - 19.99 Low
20 - 24V 200 - 240mV 20.0 - 24.0 High

This is achieved by using analog switches driven by the unused original auto range circuit. The reference and others functions are electronically inhibited by cuts and straps on the display PCB.

Upgrading the intensity display:

Adding 1 digit of precision (1mA despite of 10mA) is achieved by keeping (of course!) the reference tension of 2V of the Amps display and multiplying by 10 the tensions to measure issued from shunts resistors. This is not a problem for the + and -20V sources who gave around 1V for 0.5A output power. Originally divided by 20, it give 50mV @0.5A to the display. Then simply divided by 2 despite of 20.

But for the 6V source, the shunt resistor give 500mV @2.5A (limited power of 1.25W on the shunt). Originally divided by 2, it give 250mV @2.5A. Then, how to multiply by 10?

The solution is to amplify the voltage issued from the shunt. I wasn't sure it will be very stable due to noise in amplification. But finally, it gave good results with perfect stability, using a single precision Op Amp.

The second stage of the modification deals with the ranges.

In + or -20V, the limitation is 500mA. So, the display is:

Amps to display Divided voltage range Display
0 - 500mA 0 - 500mV 0.000 to 0.500

In 6V, the limitation is 2.5A, then displaying 2.500! As the limit of the display is 1.999, it is necessary to build now an auto-range display. It is relatively simple, thanks to the datasheet of the TC14433.

As for the voltage display in which the auto range circuit switches between dividers, the intensity display switches between amplifier gain (1 or 10).

Amps to display Adapted voltage range Display (Vref=2V) Auto Range
0 - 1.999mA 0 - 1999mV 0.000 - 1.999 Low (G=10)
2.00 - 2.50A 200 - 250mV 2.00 - 2.50 High (G=1)

Replaces 2 Voltages settings 1 turn potentiometers by 10 turns' potentiometers:

No comments. Just choose the goods ones: 10K square, not round (unfortunately the most popular). It works very well and it is quite comfortable.




The challenge of the modification is to be reversible easier as possible. I have noticed and kept the resistors I've changed to return if any to the original product (but now I won't).

The modifications consists in changing resistors values, a few straps and cuts on the base PCB and on the display PCB without changing and dismounting complex parts as ICs.

The added components are mounted on a separate board. Mine is made by wiring on a prototype circuit board. But with more time, why not a real PCB (with SMD devices to make it smaller). I used a 6 pins small connector, so the added PCB is quite simple to be disconnected from base PCB to work on (but not essential).

Components are classics ones. Only 1 Op amp, 1 quadruple analog switches and 1 logic IC. Then a few resistors and trimmers.

As it takes place in rear of display's PCB just near the power button, I've suppressed all unused pads on this added card who are near unprotected and non-isolated bare mains wires. I've drilled 2 holes for the 2 Volts reference trimmers of display board, but only one is useful now because of the suppression of 200mV reference.

The new board in situation

Schematics of the new board in situation

Board Components list:

  • 1 x TLE2021
  • 1 x DG413
  • 1 x 74HC10
  • 2 x resistor 200 ohms
  • 2 x resistor 820 ohms
  • 2 x resistor 9.53 Kohms
  • 1 x resistor 10 Kohms
  • 1 x resistor 52.3 Kohms
  • 1 x resistor 432 Kohms
  • 2 x trimmer 470 ohms
  • 2 x trimmer 1 Kohms
  • 1 x trimmer 4.7 Kohms
  • 1 x trimmer 47 Kohms
  • 1 x capacitor 0.1µF
  • 1 x connector 6 pts (M+F)
  • 14 x tests points

If you think that it's a big mess in this wiring, pics were taken at development time. Now, I've organized it.

Top side of new board

Bottom side of new board

Final mount on display

I used a good 50000pts multimeter for calibration. After settings, linearity is good, precision and stability are excellent over the full ranges of Volts and Amps displays on the 3 outputs. Now, it's perfect for me!

10V driving a 100 ohms resistor (here in fact 100.2 ohms cause of 5% resistor), giving 100mA on 3 digits now.