Lambda LQD Power Supply Repair and Modification
The Blog for this project

The Schematics, PCB files, and Simulation models are here

Introduction

The Lambda LPD and LQD lab power supplies were extremely popular in the '80s and '90s. As a result, there are thousands of these either in use, on shelves, or on the used market. At Teradyne in 2007 we had literally a pile of the LPD supplies. Lambda cleverly designed them such that the same or similar components were re-used for 1-3 channels, and from voltage and current combinations from 10V at 5A 20V, 40V, 60V, 120V, and 250V. The LPD was the original (beige) series which uses two analog meters per channel, one for voltage and one for current. The LQD is the digital (dark grey) version, and uses a single switchable LED digital meter for voltage and current.

lpd

lqd

These are quality power supplies, but the years have taken their toll on them. Fortunately they are easy to work on and in most cases easy to upgrade and repair.

Problems:

The LPD supplies have been very reliable. I had an LPD-421a-fm, 20V 2A, 2 channel supply on my bench for many years. I wanted the 40V, 1A digital display version, LQD-422. I bought a broken one on Ebay figuring I would save some money, and it would be fun to fix.

When you look at the schematics for the LPDs and LQDs, there are 2 big differences. The LPD is an all-transistor design, and the LQD uses a proprietary Lambda control IC. The other difference is that the LQD digital displays require an additional display PC board, and also require separate, isolated +/- 5V power supplies to power the digital meters. This power supply is provided by an additional transformer winding, diodes, filter caps and +/- 5V regulators on the main board. Here are the schematics (also in the LPD and LQD service manuals) for one channel each, of the LPD and LQD.

lpd schlqd

The LQD schematic shows several Lambda proprietary part numbers for key components. Don't panic, most of these are Lambda versions of pretty common ICs.
IC401, FBT-00-128 is Lambda proprietary. There is no substitute. You may be able to buy one on Ebay, or pull one from a scrapped power supply.  Fortunately they are fairly reliable.

By the way, in the late 70's and 80's, Lambda was a semiconductor provider and sold several of these power ICs, particularly 3-terminal regulators in TO-3 packages.

Models:

Here are the LQD 2 channel models:
Here are the LPD 2 channel models: Notice a pattern?

Troubleshooting:

I'll describe my experience repairing a LQD-422. It had several problems which are fairly common. The two power supplies would output correct voltages, indicating that the main regulators were working, but the 2 digital meters did not work at all. I removed the cover and the screws that hold one of the main boards. The board unfolds nicely and can lay flat on the bench.  I measured the raw +5V and -5V supplies with a scope and saw many volts of 120Hz ripple, indicating bad (open) filter capacitors. I noticed that the filter capacitors C407, C408 and C409 were bad and replaced them all on both channels. Some of these capacitors had red plastic caps that had actually fallen off.  I noticed that a few other caps had already been replaced. I replaced all the original capacitors that had red caps with electrolytics I had laying around. Feel free to use replacement caps with the same or higher voltage or capacitance, the values are not critical.  After replacing the caps the digital readouts worked, but.....

Panel meter design problem:

The displays are working, kind-of. The output of supply 1 is set to 0.0V, but the readout indicates 1.0V. Also one segment is missing on the Tens digit of channel 2. Also one knob broke so a substitute is installed.

busted


There were significant offset errors on both channels digital meters. With the voltage set to 0.0V or the current at 0A (no load)  the meters read 8-10 counts. This is significant since 10 current counts is 100mA, and 5 voltage counts is 1.0V. These are big errors for any power supply.

I tried hacking in an offset trim consisting of high value resistors to correct for the errors. It worked, but something didn't feel right about this 'fix'. Also needed 2 or 4 trims for the 2 meters and 2 ranges. Gross.

I looked at every photo of every LQD supply on Ebay. Some of them had the meter switch shown in the AMPS position. And sure enough, many showed 5 to 12 counts of offset error, just like mine! Obviously a worldwide conspiracy.

What is the source of this offset? Good question. I measured the input to the ADC, it was 0V. So the ADC was causing the error.  Was it the old ceramic ICL7107 chip? Was it the board? I bought a cheap Asian panel-meter board that used the same ICL7107. The kit that cost $8. I built it, and it worked fine, no offset error. So I bit the bullet and un-soldered one of the 40 pin ICs from the LED board, installed a socket, and replaced the IC with the one from the Asian board. Same offset problem. So it isn't the input, and it isn't the chip. Then what the heck is it !?!. I examined the circuit for the display and the recommended circuit for the ICL7107, and found a difference. The schematic (above) shows pin 35, the REF LO pin  connected to the +/- 5V return, also the input return. Effectively COM for this meter circuit. But notice the note that says -2.5V. The actual board measures -2.5V on pin 35. So what's it gonna be, buddy!! COM or -2.5V?

I looked at the various ICL7107 / 7106 manufacturers data sheets for recommendations regarding the use of REF LO (pin 35) and REF HI (pin 36). Here's what Renesas says. Notice that they show the REF pins biased to V+, or to COM, but not V- or -2.5V.
7107ref

I rewired the display board to connect pin 35 to COM, similar to FIGURE 4B, except that Lambda uses a 2.5V reference (TL431).  This took 3 cuts and 3 wires. The offset error disappeared, Hooray! I rewired the other channel, same good results. I suspect that Lambda fixed this problem on later units, and that the schematic in the service manual wasn't updated correctly. I don't have a newer unit to prove this out.

Photo of rework

Schematic of rework
rework

While working on it, other things broke:-(

To better measure the electrolytic C35, I un-soldered one pin of the voltage setting 10 turn potentiometer R31. In the process,  the pin broke off the pot. Unfortunately R31 on the 40V supply is 40K ohms. On each power supply the pot value the output voltage x 1K, so 20V is 20K, 40V is 40K, etc.  Good luck finding a 40K, 60K or 120K 10-turn pot anywhere. I bought the closest one I could get, 50K, planning to add a parallel resistor to bring the resistance down to 40K. There is a place on the PC board for the parallel resistor.  I ordered and installed a 50K pot and found that the 40V supply had enough headroom to go to 50V, even at 1A load. Hey, I got 10 extra volts on one channel for free! Notice the empty slot on the back of the old pot where a pin should be.

pot

When I first operated the supply, the 7-segment LEDs were fine. But after a while I noticed one segment was dead. Unfortunately these LEDs are hard to find. I finally located an HP substitute,
HP 5082-7731 on Ebay and bought 5. It worked OK, but is dimmer than the original.

The original 4 knobs on the LQD look nice, but they are plastic, and two of them were broken. The 1-turn pots on the current ranges are hard to turn, causing the old knobs to crack. I added some more rugged but mismatched knobs. I'm on the lookout for 4 matching ~1" knobs. Or not.

I also noticed that one of the status LEDs for AMPS on channel 2 was out. I replaced the LED with one I had.

Update 2/17/21

These fixes were originally made in Sept 2019.  As of early 2021, this supply has been in use nearly every day for over 2 years with no problems. 


The Blog for this project
  Dave's Home Page

Last Updated: 2/19/2021