The previous comment re: insulation is due to the fact that the wire is multi strand & you have to get the insulation off of all of it, or the resistance will go up.

If you did an IF alignment & the cap was seriously faulty the faulty side may not align.

What has not been detirmined before, the shotgun was used, is in which stage the fault is.

Is the voltage on B+ being monitored? This is more the relm of an analogue meter fitted with IC clips.

If the voltage lowers it is likely a short or major leak. If it rises something is going high, or a valve is shutting off. A grid slowly going more negative due to a faulty grid leak will shut off. Positive getting into the AGC or onto a grid via a leaky cap (even new fail) will stop a valve (even damage it).

That is a trap circuit as it is backbiased, C27 should not ground to chassis and C32 is positive to chassis. It will destroy the bias if it is the wrong way around and the set will stop. That also ruins the cap.

Fault finding starts with the Power supply, then the amplifier, then the RF stages.. In a backward direction.

Addendum: Philips 123S circuit is labeled differently

C25 is back bias cap positive to chassis Neg to CT C11 & 24 suspicion: C11 can feed positive into the AGC. enough set stops (valves can be damged)

Back bias resistors need checking. Voltage across them can be a guide. All current passes through them, any change affects bias.

Marc

Since I had the IF transformer open, I carefully scraped away the wax from the two caps. For the fine work I used a dental pick that I find very useful for vintage computer work. Both caps are standard micas, marked as 100pf. I had one ceramic 100pf to hand, so replaced the one on the primary side. In the process I re-soldered the coil leads and both coils now read to spec at 12ohms. I don't think the coils have moved at all in relation to the cores - the wax there has not been touched.

However on testing (TX still outside its can, attached with test leads) I get no tuned signal at all.

Decided to take Marc's advice and check all valve voltages.

On V1 (Frequency converter) the screen voltage is 17 (spec 20).

On V2 (IF amp) screen voltage is 54 (spec 45).

On V3 (audio amp) screen is 18 (spec 9). Also plate is 100v (spec 10v).

I checked the pull-down resistors that set the plate voltages.

For V3 screen, R13 reads 300k (spec 1meg).
For V3 plate, R14 reads 200k (spec 250k)
For V3 plate to centre tap (via cap 21), R15 reads 1.4meg (spec 1meg)


For V1 screen, R22 reads 250k (spec 150k)

I don't know how significant these differences are, but I plan to swap all those resistors as well as any remaining mica caps. The main power caps 20, 22, and 25 were replaced recently and bias voltage seems to be OK.

Maven

If these mica caps are actual caps with two leads and encapsulated in that reddish brown hard goop, then I'd doubt that they had failed. The cap on the primary side will have B+ on it, but it also has nearly the same B+ voltage on the other side of it. The two caps in the Ktran transformers were formed on a single sheet of mica, and problems occur as there are trace amounts of silver bridging the two caps side by side.

The coils are usually made with Litz wire, which is multiple insulated strands bundled together, like stranded wire. If a connection on any of the strands has broken, and makes and breaks randomly, you'd be varying the resistance and Q of the tuned circuit, which may make noise. The insulation on the wire is designed to burn off at soldering temperatures, so you may want to try touching with a glob of solder the wire next to the terminals, which hopefully will solder all the strands together.

I now think that there was definitely an intermittent break in the internal wire at lug 3, which would explain volume varying when external pressure was put on that lug. Probably some ham-fisted soldering by yours truly is to blame, combined with the fact that each time heat is applied to the lugs, the internal beeswax melts and would potentially flow into any gap, say between strands.

Rebuilding those connections was quite difficult and getting the tiny hair-like strands clean was a challenge - it took some dabbing with flux, heating, and scraping, to get firm tinning, before repairing with some extra lengths of three strands of fine copper.

Meanwhile I am still trying to get all B+ voltages correct, replacing several resistors and while I am at it replacing some mini valve sockets with ceramics. Not yet able to test repaired IF transformer apart from confirming expected resistance across the coils.

Maven

Jaycar & others sell a binocular type magnifier that sits like an old fashioned tennis shade. These can be used with corrected glasses & maintain depth of field. Ideal for fine stuff.

The idea for stress relief & wire being too short is to use a bit of offcut from a cap or resistor to make a post. Then hook to that.

I have only ever seen one mechanical fail in an IF cap in a locally built set in over 40 years.

Most resistors in circuit will read wrongly. If they read high in circuit, they will be high out of circuit. Plate & grid resistors are the lest likely to give a wrong in circuit reaading. resistors with good electrolytics across them will not read correctly.

The Screens may only have one B+ resistor, not always two. Carbon resistors tend to go either open , or high, rarely low. Check the resistors as you change caps, especially if one end is out of circuit. That also saves rework & frustration later.

I fix commercially 10% out for a resistor equals replace.The meter that you use can affect the reading of the voltages +/- 20% is considered normal variance for voltages.

Marc

Why no signal, wrong voltages?

I have the IF transformer back in circuit but am getting zero tuned signal.

The IF coils both show resistance at spec 12ohms, and the internal caps measured OK.

The rectifier is putting out 238vdc onto the positive rail.
Bias seems to be correct at around -6v.

On the IF Txf the primary coil reads 238v to chassis ground. The secondary coil reads nothing with a digital meter, but with oscilloscope I can see a sine wave between zero and -10V, which flattens to 0 when I ground the scope to chassis.

Every cap attached to IF txf or to V1, the 6AN7A, has been replaced, and resistors are tested or replaced to be on spec. I've swapped the 6AN7A and also the 6N8 at V2 with no difference.

The biggest voltage discrepancy is on V3, the second 6N8. I've replaced the resistor to spec (1megohm) but am still measuring 100v at the screen, where the spec says it should be 10v. Plate voltage is also double spec, at 18volts. I don't get it.

I think the audio stages are OK because I get the familiar loud 50Hz buzz if I touch the PU input, and the hum is filtered by the tone control capacitors.

Maven

I really need to see a circuit for this. I have issues with the voltages on the 6N8. What are you using to measure it.

If the B+ rail is running high a valve is likely not conducting.

The det audio 6N8 is supposed the be providing the voltage drive (swing) for the oudio output. That valve normally has a seemingly high value cathode resistor, to achieve this.

You fix these things going backward. Make sure of power then the AF and then the RF.

The secondary of an IF normally will only have the grid voltage on it.

I am expecting a wiring error on the V3 screen if it is like the 132L that 1M resistor on the screen (pin 1) is liable to have a 250K resistor as the earth leg of a voltage divider. It will possibly go to chassis via the speaker transformer secondary.

If this is the way it is? The resistor is probably open, missing, or someone removed the neg feedback wire.

Re examine the circuit and wiring of that valve

Marc

What are you using to measure the voltages?

The voltage chart for the Philips 123 has this important note under it:

"These voltages are measured with an 1000 ohms per volt meter."

This is a very low impedance meter that loads down the circuits being measured quite a bit. These type meters require 1mA for full scale deflection so if you do the maths (ohm's law) measuring V3's screen which is fed via a 1 Mohm resistor, 10v is about what you would expect.

If you go and use a modern digital multimeter with a typical input impedence of 1M to 10M, these place a much lower load on the circuit being measured and so you will measure a much higher voltage than that shown in the manual.

Good point about measurement error. I have been using a DMM. The circuit is incomplete at the moment as I am waiting for a couple more capacitors.

Meantime, I found a possible short across the variable tuning capacitor (bent fin), which might explain the complete loss of signal.

I 'll check all these again once I have the components, and I will use an older analogue voltmeter at that point to measure the sensitive voltages.

Is it practical to put a resistor in series with a DMM? I could hook up a 2K potentiometer and see what happens to the DMM readings when I adjust it.

Maven

Yours is not the first case of not understanding the ramifications of using different meters than those of the era and often as specified on the schematic.

Analogue meters draw current from the circuit and present a loading to the circuit that changes with their ohms per volt and the range that they are on.

1000 ohms per volt is read as at " full scale deflection". eg a 10V range at fsd is 10 x 1000 = 10K load.

Digital's normally use a voltage divider system and the divider presents a constant load of say 10M irrespective of the range.

Some digitals can be confused by RF riding on DC and can be useless with dirty (brush motors & generators) and fluctuating voltage.

Marc

The use of a DMM in lieu of the 1000 ohm per volt meter specified can cause a higher reading than expected, as Marcc has noted.

You can fudge the DMM to look like a lower resistance meter by putting a resistor across the DMM leads, the resistor being calculated as Marcc specified.

In the above case, on the 10V range, the resistor is 10K, in reality in parallel with the 10M meter input, but close enough and well under the meter accuracy anyway.

Of course, remember to change the resistor for another range, 100K for the 100V range, etc.

Harold

So with maximum voltages in the circuit of <300v, it looks like it could be useful to make up a DMM probe extender with a 500k potentiometer or trimpot on it, so as to adjust a load to the appropriate range for the particular test point.

I don't quite understand why two valves of the same type (in my case, 6N8s) would have substantially different voltages on the plate. One is the IF amp and demodulator, the other is the audio amp, before the 6M5 power amp.

Maven

It's all down to the application. V2 is the IF amp and needs to operate with as much gain as possible so it is run with a high plate voltage.

V3 is used as an audio amp and the gain needs to be fairly low so it is running on a fairly low plate voltage.

You'll notice that the V2 plate connects to the HT via the IF transformer with no added resistors in series. V3's plate is fed via a 250 000 ohm resistor making it very sensitive to changes in current.

I've replaced almost all suspect caps and been able to do more tests.

The audio and power stages work OK - I can attach an external tuner to the Pick-Up terminals and get full performance.

I am suspicious that the oscillator coil or V1 has a problem generating the correct IF.

My DMM has a Hz meter. When I test between the output pin of the oscillator coil and chassis ground, the DMM shows me around 1400kHz, with some move toward about 1200kHz if I move the main tuning capacitor- surely it should be 455kHz? It seems to be about 1MHz too high.

The caps and resistors attached to the oscillator coil are all new, as are those attached to V1, the 6AN7.

Any suggestion where to look? I've checked pin assignments multiple times, but something is clearly wrong here...

Pins on the valves are numbered clockwise, looking into the pan, right? When I replaced the old socket for a new ceramic one, I resoldered pin for pin, one at a time.

Maven

I think it is behaving as expected from your tests.
With an IF frequency of 455 kHz, the local oscillator runs at either 455 kHz below or above the received frequency. The sum/difference in frequencies when mixed is what produces the 455 kHz IF frequency, not the local oscillator.

Your radio is oscillating at frequencies 455 kHz above the desired receiving frequency so it looks like the oscillator section is working in some fashion.

You could have problems in the mixer section or a number of other places. The amplifier stages are at least tested as working. You need to do the same with the RF/IF sections. To do this you really need a 455 kHz oscillator preferably modulated to confirm that a signal can get through the radio.

You can also use another AM radio to listen for the local oscillator of the radio you are trying to fix. Set the working radio to around 1500 kHz, then try tuning the radio you are fixing around the 1050 kHz area. At some point the working radio should go quiet as it picks up the local oscillator signal of the other radio.

To solve this, you really need to be methodical and logical. Measure the voltages on all the pins of the Mixer/Osc valve and the IF/Detector valve with a digital multimeter and post your results here. Don't worry about whether they match the values given in the circuit diagram, as we have already established that the factory figures are not very meaningful nowadays.