Friday, May 6, 2022

E36xx Family of Power supplies


This is a collection of notes and information on how to repair these supplies. I happen to have and use a number of these power supplies from various vintages. Some of them were bought in working condition, some of them as-is. Some of them developed issues over the years  and i had to repair them. 

These supplies are extremely long life. they were originally designed in the late 80's and some models are still in production in 2022 today with Keysight badging !

Overview

The original E36xx range consists of 9 supplies.


E3610A / E3611A / E3612A

Dual range , single output, CV and CC mode. Adjustable current limit

Note : These supplies are region specific and either 100, 120 or 240 volt. They are NOT switchable

E3614A / E3615A / E3616A / E361A
Single range, single output, CV and CC mode. Remote programming (voltage or resistance), remote sensing , rear outputs, tracking ability, adjustable overvoltage protection. Adjustable current limit
Note : These supplies are switchable between 110v and 220v
E3620A
Isolated dual output, CV and Cl mode. Fixed current limit

E3630A
Triple output, second channel has positive and negative rail ideal for split supply opamp circuits.

CV CL and tracking mode. Fixed current limit.

Specifications

More details can be found in the family specification datasheet

E36xxA Specifications

E3610A / E3611A / E3612A 

These power supplies only differ in terms of the maximum output voltage and current. They share a common schematic and changes are purely based off component values/types.

Principle of operation

These power supplies use a simple but very effective means of regulating both voltage and current. In fact, the base design is used almost unchanged in even the latest bench supplies of the E36 series.

At first sight the regulation mechanism looks strange, but is in fact very simple. Two "helper" voltages (+12V and -12V) are superimposed on the output. 

The power transistors are driven into conduction through a resistor R3 (in case of the E3612A the power transistors are MOSFET type and an additional bit of circuitry is required). A sink circuit in the form of Q2 can deviate the current from R3 towards the -12 volt.

This transistor is under control of two regulating loops. 

Opamp U1 monitors the output voltage. if the voltage reaches setpoint its output will commandeer Q3 to lower the conduction in the power transistors thus limiting the voltage.

Opamp U4B monitors the voltage drop across R2 and its output can commandeer Q3 to conduct more in case of overcurrent. The net effect is that the power transistor is pinched of thus lowering the output voltage.

The output of the two opamps is "analogically" OR-ed using CR4 and CR5. Any "error" reduces the output voltage by making Q3 pull the base of the power transistors down harder.

Two comparators (U5A and U5B) monitor the output of the two error amplifiers and turn on the LEDs indicating CV (constant voltage) or CC (Constant Current) mode.

U4A is responsible to make a reference voltage for the current control loop. The classic way is to short the output and simply show the actual short circuit current. The problem is that, when doing this, the output voltage goes to zero. This may be undesirable. So they opted in this design to use a control voltage to set the current limit. The relation between this control voltage and the current is a constant and thus linear. All that is required is to "translate" the control voltage to a current to show on the display. Switch S1A toggles between the Setpoint and the actual value. The German engineers have a nice denomination for this : "Soll/Ist wert", translated :  Should be/Is value. 

Power stage oddities

Depending on the vintage of the schematics they contain mistakes or oddities. The schematic is drawn for a E3612A model and shows the IRF440 MOSFETs and additional circuitry. For the 3610 and 3611 both these transistors are in reality biased Darlington types with a built in reverse diode as well.

E3612A has C22, C23, R38, R39, R43, R44, R45, Q4 and Q5 installed to drive the power MOSFETs.
E3610A and E3611A use biased Darlington Transistors. These cannot be substituted with regular Darlingtons

Note : The ON Semiconductor datasheet shows the PNP version (2N6053 / 2N6054) of the transistor. In reality the part is a NPN version.



R1 = 8K
R2 = 50 Ohms

The internal biasing resistors make this transistor hard to substitute since you have no access to the emitter/base node to roll your own.
In a pinch you can put the 2N6055. It is essentially the same part but with a lower breakdown voltage. (Vceo is 60V instead of 80V). This has no impact on the power supply except its ability to hold external overvoltage. It has no impact on the E3612A since that one uses the IRF440 mosfet anyway.

Preventive repair

The capacitors in this supply are under a lot of thermal stress. The used series suffers from seal degradation around the pins and over the years the capacitors start leaking and drying out. The leakage can damage the printed circuit board and dissolve the copper traces while a dried out capacitor loses value and the ripple will increase, especially under load.

Recommended modifications

  • It is recommended to replace ALL electrolytic capacitors, EXCEPT the bulk capacitor. The bulk capacitor is a snap-in type and uses a different seal type around the pins that does not degrade.
  • Remove C5 and C4. they are disc capacitors and can couple noise into the output
  • Since we will change capacitors : change C12 into a 470uf 50v capacitor (was a 330uf 35V)
  • Replace R10 with a 21.5 Ohm 1/4W 1% metal film resistor
  • Replace C10 with a 6u8 35 Tantalum, for example AVX-Kyocera TAP685K035S
  • Install a bleed resistor (R46 in newer schematic) across the output. The value depends on the power supply model. The resistor is installed between TP3 and TP4. The board layout actually has the perfect spot to drop this in place.
  • Under certain conditions (capacitive loads) it is possible to drive the supply into oscillation. This can be mitigated by placing a small capacitor in parallel with R6 (200..300pF) to create a frequency roll-off in the gain of the system. 

Capacitor replacement

The original capacitor series used is long obsolete but replacements do exist that are pin-pitch compatible. You want to look for 105degree rated capacitors with long life span (5000 or more hours)
Further down in this guide is a substitution list with suitable capacitors.

Missing button caps / broken switches

You will frequently find units on Ebay that have missing button caps. As these are custom made for HAK (HP/Agilent/Keysight) they are difficult to obtain. 

The closest part that is a standard product is a C&K F-series keycap. Model F1304 is the light gray variant. Other colors are F1301 (Black) , F1302 (White) and F1303 (red).
Other options are E-switch TADGRY caps

If all else fails you can always 3d-print your own. HAK has a 3D model ( STEP file) for the E3620A available on their website. This supply uses the same button for power. Simply open in your favorite 3D modeling program and extract the button. 

The smaller square button used on different models can be replace by a C&K F21 series keycap

The power and range switches are ITT Schadow NE18 series. ITT Schadow has been absorbed in the C&K family and these switches are still available today. (see also the substitution section below.)

Corrective Repair

First perform all preventative repair, irrespective of problems, before diving into this section.

These power supplies are very robust and are hard to damage. Apart from the leaky capacitors there is very little that can go wrong and troubleshooting is very easy. Assuming the helper voltages are ok : If one of the control loop misbehaves it is one of the opamps that has failed, or, in case of current : the sense resistor. If both loops failed then the issue could be with Q3 or the power stage.

The low cost and low amount of parts that can be damaged makes it so that, worst case, you simply replace all the actives and diodes. You will be barely out of 10$ to replace all semiconductors (barring the two power transistors). There are no unobtanium parts used anywhere.

Cleaning up after leaky Capacitors

Assuming radial capacitors : A capacitor that has leaked its electrolyte will fail in one of a number of possible ways. If it opened at the top the electrolyte will try out and the capacitor will decrease in value. Very little other harm will be done.
If the part leaks at the bottom the net result for the capacitor is the same : it will decrease in value, but other damage will be done to the circuitry around.

What is Electrolyte ?

Electrolyte is a conductive liquid that forms one of the plates of an electrolytic capacitor. The key properties are
- conductive
- able to reform the aluminum oxide (Al2O3) dielectric of the positive plate.
- be chemically stable with low vapor pressure and inert towards the sealing material and pin material.

There are various formulations of electrolyte and the real composition is a closely guarded secret. In general there are three types

  • Ethylene glycol or Boric acid based : medium to high voltage capacitors up to 85degree C. contain about 20% water.
  • Organic electrolytes such as Dimethylformamide (DMF), γ-Butyrolactone (GBL), and Dimethylacetamide (DMA). These are used in wide operating temperature parts up to 150°C. They have good long term properties and low leakage. They contain virtually no water.
  • Water based electrolytes. Up to 80% water means it can carry a lot of salt ions to increase conductivity. This gives very low ESR and the ability to handle high ripple currents.
Water based electrolytes got a bad rap in the early 2000's due to lots of capacitors bursting and causing mayor failures in lots of products. The root cause was a stolen , incomplete formulation that found its way in lots of capacitors. The inhibitors used to prevent the hydration of the aluminum were missing or incorrectly formulated. The hydration process creates heat and gas and burst the pressure vent at the top of the capacitors. the capacitors dried out and the circuits failed.

Today almost all long life , low ESR , high temperature parts are water based. they have service lives of up to 10.000 hours at 105°C.

There are other types where a polymer is used but this is beyond what we are dealing with here.

Cleanup on isle capacitors 

In order to clean up the leaked electrolyte we need to know what type is used, therefore we need to correctly identify the capacitor series.

Component level troubleshooting

If you want to do component level troubleshooting without using the shotgun approach. Below is a walk through of what to check.

When performing measurements pay attention to where you place the negative probe of your meter. These supplies have multiple power domains and what is "ground" for one domain is not necessarily "ground" for another domain.

  • Reverse voltage damage : in case a reverse polarity voltage has been applied externally the diode CD3 will most likely be damaged. It can fail open or shorted.
  • Verify the power rails. Leaking capacitors can create enough of a load to blow the fuses F1 F2 and F3. These are 1 ampere solder-in Littelfuse parts.

    1. Measure between TP9 and TP10 to check the 5 volts for the displays. Check F3 if failing.
    2. Check the Reference supply. 
      • Measure between TP6 (GND) and TP1 for +12 volts
      • Measure between TP6 (GND) and TP5 for +5 volt
      • If both are missing : Fuse F2 is suspect.
      • If +5volt is incorrect U3 may be broken (LM336BZ-5.0). Pay attention when ordering this part. There is a 2.5 volt version that is much more prevalent than the 5 volt version, especially eBay sellers typically only have the 2.5 volt version.
      • Measure between TP6(GND) and TP8 for -12 volt. If missing F1 is suspect.
    3. Main rail : 
      • Check between TP4 (GND) and TP2 for the unregulated voltage. The voltage depends on the position of the range switch (S2A). S2 gets a pretty large "hit" when switching from low range to high range as C2 needs to be charged up. I have seen cases where S2 developed burned-in contacts.
    4. No output voltage 
      • Turn both voltage and current limit pots somewhere halfway.
      • Check R35 and R36. Their value is dependent on model !
      • Check that the node R4/R5 is not at, or close to, -12 volt. Measure from TP8(GND) to the R4/R5 node. You should be able to move the voltage up and down using the voltage adjust potentiometer.
        • If the voltage is close to 0 volts (relative to TP8) the current error amplifier is most likely damaged.(make sure the current limit pot is somewhere halfway).
          1. Check the Current sense resistor R2 and both input clamp diodes CR10 and CR11.
          2. Check Cref. Measure between TP6 and TP7 and verify the current limit voltage can be adjusted with R19. Check potentiometer R19 (Front panel) and its wiring.
          3. Replace U4 if nothing else works.
        • Check the R37 (front panel voltage potentiometer) operates properly.
          1. Check diode CR6
          2. Check diodes CR4 and CR5
          3. Replace U1 if all else fails.
      • If the voltage at R4/R5 moves then the transistor Q2 or the power transistors are suspect.

Difficult to identify parts

Power resistors (wire wound)

Power resistors ( Wire wound ) are made by a Korean company HMRohm (not to be confused with Rohm from japan). www.hmrohm.co.kr. The ARW03 (3watt) and ARW05 (5 watts) are used.

These are high temperature Silicone covered  Alumina substrate Nichrome alloy resistors. 

The 3W versions can be substituted by Vishay AC03 series. For example the 390 Ohm bleed resistor can be a AC03000003900JAC00

The 5W current sense resistor can be replace by Vishay ALSR05 series.
E3610A : 0.1 Ohm,  ALSR05R1000FE12
E3611A : 0.2Ohm, ALSR05R2000FE12
E3612A : 1.78 Ohm -> ALSR05R2R000FE12. NOTE : current realignment will be necessary !

Power and range switch

These were originally made by ITT Schadow division which is now a part of C&K.
The most easily obtainable is the NE1838EE and this part will fit both the power AND range switch on these supplies. The power switch is technically a DPST version but the PCB has the holes for the two extra pins in the DPDT variant.

NE182UEE6AMP is the correct part number for these. There are other variants but they either do not have the solder lugs on top or have additional PCB mounting clips that interfere with the board (and these are longer so they don't fit the front panel) . Pay attention when buying these off eBay as there are a lot of flavors.





Current set switch

This to is an ITT Schadow design now part of the C&K family. The switch is an F-series, notably the SF2UEE. Any of the following part numbers will work : F2UEE, F2UEEAU, F2UEE TB or SF2UEE.

Solder-In Fuses


The three 1A 125V fuses are LittelFuse PICO 251 fuses. Model 0251001.MXL 

Component Datasheets

The following is a collection of datasheets for key components used in these power supplies. Some of these components are obsolete and the datasheets are sometimes hard to find.
The original documentation does not really list the manufacturers of the parts.


Component

Description

Function

MJ3042 Transistor

 

 

IRF440 Transistor

 

 

Component Substitutions

As the original parts are going, or have gone, obsolete they need to be replaced with different but compatible parts.

Component

Value

New Part number

C2

Nichicon LGY series 10.000uF 50V

LGY1H103MELC40

C3, C7, C13

Nichicon UPJ Series 470uF 50V

UPJ1H471MHD6TN

C17

Nichicon UPJ Series 4700uF 25V

UPJ1E472MHD

C9, C15, C19

Rubycon YXJ series

50YXJ1M5X11


Note : This part list includes the upgrade for the 330uF capacitor to a 470uF version

Service Manuals

The following is a collection of service documentation available from HP / Agilent / Keysight. There are various revisions of these documents and some have more or different documentation than others.

File

Models

Comment

 E3610 Service Manual (old)

 E3610 / E3611 / E3612

Old schematic without bleed resistors. Power transistor section is wrong (shows two MOSFET instead of Darlington)

 E3610 Service Manual (new)

 E3610 / E3611 / E3612

Power transistor section is wrong (shows four MOSFET instead of Darlington)

Service Bulletins

Post design release there have been a number of service bulletins that describe issues and give remedies to solve the problems in the field.. Below is the collection of these bulletins.

File

Issue

 E3610A-01

 Output terminal binding

 E3610A-02

 Voltage Transients during power-off

 E3610A-03

 Voltage overshoot

 E3610A-04

 Oscillation at output when output grounded

 E3610A-05

 Oscillation under load change

 E3610A-06

 Power socket can dislodge

 E3610A-07

 Oscillation at full load

 E3611A-01

 

 E3611A-02

 

 E3611A-03

 

 E3611A-04

 

 E3611A-05

 

 E3611A-06

 

 E3611A-07