HP 3478A with no AC Voltage or AC Current Reading

I found this nice unit on eBay at a discounted price as a part unit, because according to the seller, the unit doesn’t read AC voltage or current. Otherwise everything else seem to work perfectly.

After some researching and digging through the schematic. There are few components like K104, U102 and U301, U302 and U303 ICs possibly failed.

Before proceeding further, if you are going to make a similar repair make sure it is disconnected from the AC line and take extreme care not to short battery. Otherwise all the calibration settings in CMOS RAM will be lost and need recalibration.

Once, I received the unit, I began tracing AC signal through K104, U102 and U301, U302 and U303 input. The signal made it through and range switching seem to work just fine. However, the output of U303 (AD536A True RMS-to-DC converter) is stuck at near ground regardless of the input signal.

To confirm that U303 is dead, I cut the JM302 to DAC section and reconfirmed that AD536A is faulty. Then to confirm that meter is still able to read the AC input from the DAC. I placed a .5V at DAC side of JM302 and the meter is responding, so the DAC is able to read the converted AC signal and it is a good sign.

After some research, I finally ordered Analog Devices AD536AKH replacement for U303 from eBay. When the unit arrive, I began surgically removed the old unit by clipping the legs and cleanly removed the old solder and pins before installing the AD536 replacement unit and bridge the JM302 jumper.

The AC volt and current is working now and the reading is slightly off and need recalibration.

Next improvement is to replace the lithium battery.

IMG_0122 IMG_0132IMG_0131 IMG_0135 IMG_0139


Atmega8 Experimenter Board

I am making PCB design for ATMega8 experimenter board. So, far has all the basic stuff on a single board. Hopefully there are some interest in it that I may able to batch create bunch of PCB cheaply.

Trying to produce this board using the toner transfer method failed miserably and messy sink. ūüôā


TRS 80 Model 100 Cassette Interface Cable for PC

At the time when I was about to replace TRS 80 Model 100 NiCad Battery. I have a few programs on it that I need to make a backup of before replacing the battery.

I don’t have a null modem cable or any other means of saving the data.

So, I went a head and created a custom cable to record these backup using my laptop through the Mic input and Audacity program. With this cable, it’s also possible to record program on these cheap MP3 recorder using the Mic input as well.

Below is a drawing of the cassette interface which use a simple voltage divider to reduce the line audio to a level which is suitable for laptop Mic input.


To capture a recording, press record on Audacity and type in the following command on the laptop to begin saving the program.

save “cas:alien.ba”

To reload the program, type in the following command on the laptop and press play on Audacity.

load “cas:alien.ba”

Common Loading Problems

  • A weak, noisy recording in the recorded audio might cause loading problem. This is because most vintage cassette interface use zero crossing detection circuit to detect 0 or 1 by counting number of crossing in given period. Any noise introduced in the audio might cause false reading. In case of weak signal, increase the volume and most of the time it would fix the problem. If the recorded audio is noisy, try disconnecting the power adapter on TRS 80 Model 100 and run it on battery to reduce hum.
  • If you’re recording in stereo, make sure the channel that feed the ear input has audio. Otherwise switch the channel or record in mono instead. Alternatively connect the ear input to both left and right channel.
  • If you are still having problem with loading, use Audacity to zoom in and look at the recorded audio. Ensure that the recording is not clipping and audio signal is utilizing the entire range. Otherwise increase Mic sensitivity or adjust the resistor ratio to increase the Mic input level. Below is a decent sample of good recording.


Replacing TRS 80 Model 100 rechargable NiCad CMOS battery.

I have a TRS 80 Model 100 and while doing a routine maintenance. I noticed the internal NiCad is showing sign of crystal forming. Before allowing it to leak and destroying the PCB, I ordered a 3.6V 60mA NiCad replacement from eBay to replace it.

I was planning to remove PCB to do it properly, however after the cover was removed. It seem like it can be done without removing the PCB.

Using the 40W solder iron, I heated each leg of the NiCad battery from the top side and gently removing each leg. Then I used the solder wick to clean up the solder points and install the new battery in it place. Afterward, re-solder each leg with sufficient heat to ensure  proper bonding.

Once everything is put back together, hook it up to external power adapter and let it charge. Now unless, one of these electrolytic capacitors start failing. It should extend the life of this laptop another 5+ years.


Restoring HP 403B AC Voltmeter (10Hz – 2MHz)

The following instructions, document the process that I toke to restore the  HP 403B AC Voltmeter to working order.

Day 1

The unit was a bit dirty when USPS delivered it to my front door. The rechargeable battery pack is showing sign of corrosion.

First, I checked the meter movement and making sure that it is in good shape.

The moment is smooth and doesn’t appear to have any issue. With a 1.5V battery and a variable resistor, I checked the full-scale reading and a few other reading just to confirm the meter linearity.


A quick check of the voltage across the capacitor C21 and I am getting a steady 42.28V DC.

Next, I checked the battery +13, -6 and -13 supply voltage. I am getting +10, -6.5 and -13, this is because the battery is so old and some of the cell are dead and not evenly charged.

Even though the supply voltage are way off! The good sign is that the unit is fully working.

The battery is showing sign of corrosion and need to be replaced. So, it’s on my next to do list.


After a quick search, I found suitable replacement battery from eBay. The seller is trying to get rid of his inventory and selling a 25 packs of 3.6V 700mAh for cordless phone for only $0.99. The only catch is the $20.00 shipping.

The meter originally came with 4 x 6.0V 225mAh Nickel-Cadmium Battery but I am going to replace it with 8 x 3.6V 700mAh battery packs made for cordless phone.

The combined voltage will be slightly higher when it is fully charged, however, it doesn’t seem to pose any problem after everything is re-calibrated.


After tying them together and solder their leads in series connection to make 2 x 14.4V batteries pack.



I reconnect and the battery to the original +13, -6 and -13 leads and put everything back together.

NOTE:¬† Don’t solder the -13 lead just yet, as you will insert the amp-meter in series with it to adjust the charge rate later on.


I made some minor change to the orientation of these batteries after soldering them in place, so that the PCB will fit on top.


With the Amp meter connected in series with -13 power line, I adjust the pot to 6.5mA as suggested by  the manufacturer. You may set it to 11mA, if you plan to use it on battery regularly.

I let it fully re-charge over night before I return to work on it again.

Day 2

As I am going over the calibration procedures.

I notice that when meter is on, it never quite return to zero. It is alway 2 ticks higher.  Which kinda annoy me.

TODO: Investigate to find out which aging components causing this.

But for now, I just re-zero it with the zero adjustment screw. This will affect it accuracy a little bit. However, when checking the voltage indication on each range the reading track very closely to my HP 334A reading. So, for now it is not an issue for me at the moment.

The strange thing is that, on these HP AC Voltmeter. There is only mechanical zero adjustment. Unlike most VTVM that I had restored. Where you have to adjust  mechanical zero, AC and DC zero.

Day 3

While calibrating, I notice that on 0.01 and 0.03 range. The meter doesn’t work at all and it’s alway peaking.

It seem the meter is fully deflected when those range are selected.

At first, I though some resistors in the voltage divider network are bad. But after checking these resistors, switches and signal path. Everything checked out ok.

So, to diagnose this further. I connected the oscilloscope to take a peek at the signal at each stages.

After tracing the signal path on the diagram, the original input signal is amplified and then going into the resistor divisor network.

When I connecting the probe and observe the output at R20 and R22, There is 11mV p-p (.048us or 21MHz) voltage showing  on the scope.



In the 0.01 and 0.03 range, the original signal coming out of the voltage divider  is so small compare that with the 11mV p-p noise.

The noise is swamping the original signal and causing significant meter reading error.

NOTE: According to the manufacturer manual, this may be caused by a noisy transistor and need replacing.

But first, I wanted to try to fix it by adding a filtering capacitors.

After adding a 0.1uF 50V to +13V and -13V to ground, this seem to eliminate the noise.

To install the capacitors, I find some open space on the PC board and solder them in place.

This help to remove the noise and making 0.01 and 0.03 range working again.

IMG_9808As seen below the signal is much cleaner now.


Day 4

Spend a few hours calibrating the meter the best I can. Due to lack of a signal generator that can put out 30V 300kHz sine wave, I have to put off those calibration steps.

At the moment, the meter is track accurately from 1mV to 1V from 10Hz – 2MHz.


  • Recap
  • Finish up calibration at 300kHz


Link to Service Manual:


Display a pop message at the end of a cron job.

Traditionally, when a cron job is finished. You will get a nice email indicating of its status.

Lately, I am quite annoyed by the amount spams that I am getting.

On my Linux laptop, I had it configured with a few cron jobs running in the background monitoring other activities.

Lately, it just flooding my in-box with messages and becoming unmanageable and hard to keep up.

Since, I am only interested to know if something interesting had occurred. There must be a better way than flooding my in-box with emails.

After, some research I found that the notify-send is included with every Linux-Mint installation. It’s quite simple to use and quite customizable and fit the bill perfectly.

Below is a sample on how to use it:

¬†¬†¬†¬†¬†¬†¬†¬†¬† notify-send –icon=/usr/share/icons/Mint-X/apps/48/gnome-gnibbles.png –expire-time 30000 ‘Oppa’ ‘You do have a life after work! So, go home!’

Free Online Virus Scanner

If you’re a vintage computer collector or someone enjoy using old computer like me. Once in a while you might, in inadvertently downloaded softwares from a questionable source which may contain virus/worm.

Overtime, I have taken many precautions to avoid such problem.  But, the idea of having to buy Virus scanner software and pay yearly subscription fee to me is ridiculous.

A better solution I found is to use a free on-line virus scanning services which will use multiple virus scanners to look for potential virus or worm.

Once such service, I found is https://www.virustotal.com, you simply upload the file in question and after while you will get the scan status report. After which you can rest assure the file virus free.

%d bloggers like this: