PS30 Repair


V1.1 23rd June 2021

  1. V2.0  9th August 2021
  2. V2.1  5th September 2021
  3. V2.2  3rd October 2021
  4. V3.0 1st November 2022
  5. V3.1 19th November 2022

For a repairer, the burst-into-life of an otherwise dead unit is a syringe of adrenalin: to have it fail and to find a second latent fault is a second philip. The obvious fault and the first repair.

As solid state RF amplifiers and Amateur Transceivers designed for base staKon and mobile use became more cost effecKve, so did the need for mains powered 13.8V, high current power supplies: As I gathered surplus transceivers, so did the need for a larger current source!

Amateur Radio operators are born helpful and remain so in the interests of enhancing their hobby and the enjoyment of other like-minded people. A reflecKon on the valuable service to Australia’s freedom are the many stories of self-taught Coast-Watches spread throughout the island chains to the North of Australia in the 1940s and reported enemy Japanese posiKons on self-constructed transmiWers. I was the first person to speak to a Darwin Ham and raise the alarm of the devastaKon caused by cyclone Tracy, 0700hrs, 25 Dec 74 using amateur radio: Darwin to Port Moresby to Sydney.

A fellow amateur had acquired an Icom IC-PS30 power supply, not working and through swapping, I was able to acquire this challenge.

The IC-PS30 is a 1980s supply, worth some US$320 at the Kme and grew from the then linear technology of transforming the mains to some 20V DC then regulate it to 13.8V: the high current demanded a transformer of equivalent weight to a small boat anchor. A new player had entered the scene [and has since taken over], the switch mode power supply [SMPS]. In thumbnail, convert any input voltage/current to DC, create oscillaKon at 20Khz to 40Khz then use a much lighter transformer core to output whatever the design calls for. The drawback is the generaKon of mulKple harmonics which tend to escape from most SMPS units and become the bain of receiving weak signals. The IC- PS30 has kept the size and weight commensurate with shack-based transceivers whilst trapping the harmonics inside the case.

The case is 241*110*300 mm and the whole unit weighs 5kg. The front face has an on-off switch and an output voltage and ammeter and associated switch. The case is vented with a fan at the rear; input power, fuse and output sockets are at the rear. Inside is a dream. A HV recKfier and smoother PCB along with an indicaKng system low voltage transformer [low current], recKfier and smoother greats the case opener. This pcb is mounted on a finned aluminium cast inverted bath tube which has a removable plate base on its under with mulKple screws to form an effecKve RF shield. This sealed structure houses the 20khz RF generator and amplifier whilst trapping the annoying harmonics. Input and output cables are bypassed to reduce further unwanted signal escape. The structure takes a Kme to disassemble and photographs are required to assist reassembly. The major problem faced is finding the faults[s].



Figure 1 PS30 Oscillator, amplifier, transformer

Linear power supplies ease the fault finding process by tracing expected voltages from the “On” switch to the output voltage. When the required voltage at any stage is not detected, the fault is in this area; the SMPS is not that simple. Whilst an external sKmulus is needed to start the process, the output is then used to power the oscillator which in combinaKon with the external DC input, provides the output: a flywheel type principle. Without an output there is no feedback and the fault is difficult to isolate.

IC-PS30 Serial 04620 was disassembled and the first enthusiasKc/guesses made. There are a number of published techniques to fault find SMPS and with the guesses having failed miserably, a low value external voltage was injected into the output. The oscillator sprung to life enabling the signal to be traced up to the C class amplifier. Without the 300V power applied, the signal was expected to stop at the base of the first push-pull transistor and it did. The base resistors had failed open circuit e.g. very high resistance and given the other circuit components, the oscillator signal amplitude was similar either side of the failed resistor. A back-up check required an analogue mulK-meter to measure the resistor value and problem isolated. Much Kme was spent on this task but the analogue mulK-meter came to the rescue: two low value base feed resistors were open circuit. The resistors were replaced and with checking that appropriate separaKon was made from other disconnected parts, the PS30 burst into life. Time to address general deterioraKon.

Filtering is used at many stages so electrolyKc ESRs were first on the list: all four 470 microfarad out put capacitor ESR measured 8.3R, 12R,13R and 21R whereas the required was 0.2R. All four were replaced and upgraded from 16v to 35v. the 4700 microfarad at the output was replaced. The 20khz amplifier comprises four NPN transistors in Darlington Pair and in push-pull configuraKon along with some low value .25w resistors and wave-shaping capacitors; these components are subject to some heat stress. Recommend an upgrade for the two 22R and the four 4.7R resistors to 1W types and replacement of the two .2 microfarad capacitors. Other resistors and capacitors in this vicinity could also be stressed but measured near to published values.

The IC-PS30 has been humming on the bench for 24 hours without fault and is ready for service with HF and VHF/UHF base staKon transceivers.

Addendum [23rd June 2021]

My contribuKon to the “gioer” of the IC-PS30, the subject of this brief, was this text and a kit of parts to correct the deficiencies idenKfied: he found more faults and thus this addendum.

The circuit is divided in two: components outside the some 20khz oscillator/amplifier/transformer and those inside the sealed “bathtub”. It seems that component designaKons have been duplicated and in one case have the same values.

Outside the “bathtub” bridge recKfier, D1 provides DC to the amplifier inside the “bathtub”. In accommodaKng either 110V, R1 and R2, both 68k are bleeder resistors across the filter capacitors. These were found burnt at the contact cap ends and read in the Megaohms. They are doused in resin at manufacture which can disguise the damage and make the bits hard to remove.

The paragraphs in this brief before the Addendum heading addressed, inter alia, R1 and R2, both 68k and again blender resistors which also needed to be changed out of cauKon.

On power on, the PS30 remained stubbornly dead. A check of the bridge D1 showed no power connecKon traced to an open circuit R5, a wire wound 3 ohm resistor. This limits the iniKal current rush on turn on. When the oscillator/amplifier boots up, relay R1 is powered by the 13.8V output and shorts resistor R5. R5 can be found at the end of Relay R1 and near the diode bridge D1. R5 was repaired and the IC-PS30 burst into life.

Thanks to Michael Barret, VK3EMB for his input. Repair Sage Phase Two [9th August 2021]

ExpecKng it to end there and a lit volt/amp meter delivering power to the rig ad infinitum was false hope. The IC-PS30 worked well on low current [some 2A] Krelessly. Doubling the rig output power from 5W to 10W and both the IC-PS30 and the rig died with a characterisKc “pot” from the power supply. This Kme the smoke was trapped inside the RF shield. The IC-PS30 was swapped out and the rig burst back into life on both VHF and UHF: not the rig.

The IC-PS30 was disassembled and inside the RF cast aluminium RF shield, both 2SC2307 BJTs
[ Bipolar JuncKon Transistor] had burst their sides, one splipng apart. As well, the uPc1042C switching regulator had Vcc short to ground. Trap for some players, checking output posiKve to ground also appeared a short but given the very large output filter capacitor bank Kme to charge and the resistance increases: pays to know the circuit under test.

The IC was replaced with a socket and a new chip along with subsKtutes for the two 2SC2307 switching grunts. Time to check the rest of the circuit. Many electrolyKc capacitors in the control and filtering circuits were changed in the original fix but a mere 10uf, 25v in the IC output was checked just in case: 6uf and 150R equivalent resistance. This capacitor is on the centre tap of the switching regulator output transformer to ground where Vcc is applied. With exciKng voltage at some 20Khz added to other induced voltage being seen by the IC, the IC gave up the fight and failed.

The moral: in repairing the IC-PS30, replace C11, 10uF, 25V inside the aluminium RF shield bathtub.

On switch-on, the IC-PS30 burst into life and given that this test was completed using 7A instead of 12A max IC BJTs, when the specified replacements arrive they will be fiWed and the power supply exercised to maximum output.

I am sure there are more sensiKve points in this well-worn power supply so this saga could well be extended.

Addendum 5th September 2021

The propheKc statement that this repair saga would conKnue was spot on. In its previous life, the owner appears to have punished it and now the PS30 is fighKng back.

Humming away in the shack and delivering just two amps, the transmiWer was keyed and the current demanded was 4 A from a possible 25 A. A phfuW! Was the response and the PS lights went out. On disassembly, both transistors in the complimentary pair failed catastrophically. The visual inspecKon was similar to massive overcurrent/overvoltage. The driver transistors were checked and appeared and repeat appeared intact.

To get to the innards, the well buried bathtub shield needs to be extracted and unpacked so trial and error is an inappropriate fault finding technique. With nothing found, the innards were repacked and

on switch-on, lights and voltage were normal. The instrument light is powered by the DC output so an immediate indicaKon of proper operaKon appears.

Cycling the power switch reacKvated the lights however a 1A load was applied to the output and the lights went out; some part of the current generaKon or the output current limiKng circuit was in control. Again a disembowelment but the in circuit oscillaKon generaKng components tested serviceable: Kme to remove the complimentary output pair and their drivers.

The lesson to be thorough and document the results as they are read had to be relearnt. On removal, one driver had a base to emiWer short. This must have been present in the in-circuit test and the oversight is put down to rushed enthusiasm. With acKve components removed, the low value, low waWage surrounding resistors had all open circuited.

page4image50181536 page4image50180704

Reassembly with outputs and drivers that matched, and replacement resistors and the bath tub is ready to close up with the RF suppressing lid. Given the transistors are mounted on the bath tub heatsink, physical mounKng first then electrical connecKon to the separate pcb pins ensures minimal assembly strain.

All assemblies were repacked into the chassis, connecKons checked then rechecked and power applied. Power was leo on to illuminate infant mortality and then removed. Without load, the output voltage remained constant for some Kme, a good indicaKon.

Not wishing to take the shine from the apparently successful repair, some days passed before the IC- PS30 was put under load. I scavenged a discarded 240V AC fan heater with a linear heaKng element. This element comprised 12 by 15 ohm nichrome wire elements supported in ceramic formers. All elements at one end were connected, at the other they were connected to spade terminals which progressively provided 3.6R, 2.6R and 1.6R. At 13.8V the iC-PS30 delivered 3.833A, 5.31A and 8.6A. These loads caused consequenKal heaKng of the element and were not applied for extended periods.

ADDENDUM 1st November 2022

The IC-PS30 had been used intermiWently on VHF and UHF and has worked flawlessly for about one year: Kme to rearrange the shack!

With the AC 240V connected and live but the IC-PS30 switch off, connecKons were being made: an AC Kngle was felt between the IC-PS30 chassis and other earth grounded equipment. The trusty analogue MM read 120V AC.

Some background. World Radio history dates this power supply to “circa 1980” when Japan’s reKculated 110V ac was acKve and neutral, single phase: no earth1. The IC-PS30 has an earth connecKon on the chassis rear and an installaKon instrucKon to connect that point to ground:


To meet Australian earthing requirements a bolt holding a tag strip near the power cord-to-chassis entrance had been modified to secure the tag soldered to the incoming yellow/green earth wire. There was a problem!

This bolt was inserted through the boWom powder coated case. The hole in the case gave bolt-to- case clearance and therefore there was no incoming earth wire to chassis conKnuity. The circuit of the IC-PS30 did provide for input power spike filtering through four 1000pf high voltage capacitors from acKve and neutral to “earth”; the insulaKon around the bolt meant the chassis was not included in that earth yet two filtering capacitors did return to the chassis and two to the incoming mains earth.

The two capacitors now returning to the non-earthed chassis made that chassis semi live and thus the Kngle. Removal of all four filter capacitors eliminated the problem. They were replaced with new 1000pf 3Kv items and the chassis connected to mains earth.

ADDENDUM 19th November 2022: Version 3.1

All went well for about one week. On turn on the PS30 would start, the meter would indicate about 14V and the meter illumination, powered by 13.8V DC output, would light. Within one or two seconds, the PS30 would close down. All loads were disconnected to ensure that the over current, under voltage circuit was not in control and the same start up sequence. If I left it for a while then applied 240V AC, the one to two second come-alive sequence would repeat. If I turned it off then immediately on, the unit was dead. A good sign that some capacitive influence was playing a part.

The fault finding sequence involves separating the PCB sitting on the “bath-tub” from that inside the the “bath-tub”. All output voltages from this board checked, although not under load. The only resort is to remove the “bath tub” and disembowel it! To trap the RF spurious inside this tub, it is sealed with a plate and many screws. With the components already replaced reasonably assured, time to start with the remainder. One anti-ringing 2,200pf capacity sheltering against the transformer [heat effect?] attracted attention and was gone/replaced in a flash….well not literally. The push pull Darlington pairs fed the RF output transformer through a 3.3uF, 400Vdc capacitor. Not a common value in the spares box.

One was ordered and then wait.

In the meantime, the BJTs were removed and checked as were the low value bias resistors: All good. When I previously removed the uPC1042 IC, I installed a socket. Paid dividends! The IC was OK but the installed 3.3uF on removal and test with a megger [I have my Dad’s generator derived voltage Megger so I can apply the voltage progressively] showed resistance rising to 2M ohm and then dropping to a low value: the 3.3uF was breaking down.

The replacement 3.3uF arrived and the pcb drilled to take the lead pitch. With a false start which we will skip over, and power application through a variac, the PS30 burst into life and delivered current.

The IC-PS30 has been returned to use as I grapple with the intricacies of D-Star communications.

I hope this brief and Addendums have been of help were you to have a dead IC-PS30.

John Baker VK3JWC


1 In 2022 Japan has Type A plugs which are two pin and Type B which have a third round earth pin.


Thanks John for the interesting article. Regards Ed.