Repair of Hi-Fi Equipment

 

 A number of items arrived at the Radio Museum today for repair (July 2026)

These are as follows:

Two Leak Stereo 20 Amplifiers

Two Leak Stereo Pre-Amplifiers

Teac Open Reel Tape Deck

These had differing reported faults and from a cursory inspection they needed new components to restore them to working condition.

I'll tackle one of the amplifiers first then fix its pre-amp companion.

 You can see in the picture, above the circuit diagram (click this for an larger drawing), several points of interest. In the centre appears to be evidence either of a design error where the engineer under-rated the resistors or perhaps there are failed components elsewhere. Looking at the capacitors there must be several suspects for leakage or loss of capacitance. The main smoothing and reservoir capacitors (top left and right) will need checking as if these are in poor condition there'll be hum on the audio.

Comparing this view with the second amplifier you can see that all four resistors down the centre of the board are supposed to be 270 ohms, however those burnt ones both read about 7Kohms. Why are the resistors burnt and very high in value? Well, anyone with knowledge of valve circuitry will tell you that the capacitors providing audio from the preceeding amplifier stage might be leaky. Any leak will feed some of the HT line to the output valves control grid via the anode load resistor of the ECC83. The presence of any undesired positive voltage will raise the anode current of the EL84. This in turn will increase the current through its cathode resistor and result in excessive anode dissipation.

The cathode resistors are 270 ohm and the design voltage at the cathode is 10 volts. Hence the resistor will dissipate (10x10)/470 or about a quarter of a watt. The anode current of the EL84 should be about 20mA but if the grid voltage moved more positive the anode current would rise and the cathode resistor would soon fail.

Here's something that will really upset the purist. Those large 0.25uF capacitors are marked with the outer foil connection at the banded end. You'll note that in one pair (in both amplifiers) the foil ends are connected back-to-front.

If you look to the right of those large capacitors.... where are the two capacitors which are present on the second amplifiers circuit board? They're the cathode decoupling capacitors which should be like those two green 50uF capacitors. Of course, being in close proximity to the hot resistors they would have failed and presumably someone has clipped them off?

Those four 470 ohm resistors seen in the second amplifier are marked 5 watts (a bit of an overkill for a dissipation of a quarter of a watt !!) but clearly someone has solved the resistor overheating problem but failed to correct the reason for this. Did the last user see a problem? Unlikely, as the audio would be reasonable but the output valves would be running red hot.

In fact if one looks at the driver stages there are additional leaky capacitors. These will apply an undesired positive voltage to the control grids of V2 but what would be the result as far as a user is concerned? The driver valve would be running too much anode current and if the audio input was high, but ws still within spec, would clip the output resulting in distortion... however this scenario would require a very loud audio output and in the average listening environment would be most unlikely. This means the amplifier fault would remain undetected.

 The second amplifier whose circuit board is shown above (said to have been fine since purchased secondhand) carries a date on the chassis of 1963. I measured the value of all the resistors and found lots in really poor condition. As a matter of interest why are the EL84 cathode resistors (R14/R15) 5 watt? My guess is that many years ago the amplifier was overhauled by someone with limited knowledge of component failings. The chief reason for excessive anode current is a leaky capacitor carrying a positive voltage to the control grid of an output valve so I disconnected the four large 0.25uF capacitors (C4R/L and R6R/L)and tested them. I used a bench supply providing a maximum of only 31 volts and found that three capacitors leaked between 12 and 15uA meaning that they would have a resistance leak at 31 volts of about 2Mohms.Bearing in mind the ECC83 anodes are sitting at over 200 volts this is really bad as up to 40 volts could end up on the EL84 grids. In fact this isn't the full story because two of the four 470kohm ECC83 grid leaks (R12/R13) measured 660kohm and the other two were open circuit.... meaning that the whole anode voltage of the ECC83 is available at the control grid of each of two EL84s.

Anyway cutting to the chase the picture below shows the components that I replaced... then a view of the repaired circuit board 63 years after it was made.

Next I'll need to apply mains and see if the new components restore operation... hopefully the valves are in fair order.
 

 

 Before applying mains power I removed all the valves except the GZ34 HT rectifier and carried out a few basic checks and soon found the heater supply to the left channel valves was peculiar. After several tests I discovered there must be a dry solder joint (in the green circuit) under the circuit board but as the board is mounted in the wiring it wasn't a good option to look at the wiring, underneath it, carrying the dry joint.

The electro-mechanical design of the amplifier is intended to impart a look of symmetry and because of this one can work out the likely way the heater wiring is done. Using a little logic, the most likely place for the dry joint is close to the mains transformer where the heater supplies to the left and right channels split (the point of maximum current). This means that supplying a secondary feed from the heater winding to the nearest valve in each channel will overcome the problem. The heater winding is centre-tapped to ground leaving a green circuit and a grey circuit. I added two pairs of wires from the 6.3 volt winding to the heater pins of the nearest EL84s. I then checked that all the valve holders carried these connections.

The next step was to check the HT circuit. This was satisfactory so applied 240 volt mains. The HT rose slowly from 383 volts and eventually measured 387 volts. The heater circuit measured 6.8 volts AC.

I noticed a grommet in the end of the chassis and below this was a shorting wire in the transformer primary marked "switch" . As there's no ON/OFF switch in the amplifier I removed the grommet and added a toggle switch in the hole. This will make testing a lot easier.
 

 I refitted all the valves, connected a loudspeaker in the left channel and powered the amplifier. The thing warmed up and there was a low background hum, probably because the input was open circuit.

The following table lists the voltages measured at the valve pins. The raw HT was 329 volts (325 volts), smoothed HT was 305 volts (310 volts)and the HT at the driver was 285 volts (295 volts) with the input valve HT at 118 volts (120-195 volts).The DC voltage across the 100 ohm HT smoothing resistor was 14.4 volts indicative of an HT current of 144mA. All figures in brackets are those taken from the circuit diagram. The circuit diagram indicates that the HT current is 150mA so my figure of 144mA is pretty good. Discrepancies in voltage reading at the input stages may be due to R20 and R19 being slightly high.

 

 VALVE

 PIN 1

 PIN 2

 PIN 3

PIN 4

PIN 5

PIN 6

PIN 7

PIN 8

PIN 9

 ECC83-C

 100 (135/160)

 0

 1.0

 H

 H

 101 (135/160)

 0

1.0 

 H

 ECC83-L

 210 (215)

 22

 31 (37)

 H

 H

 208 (210)

 21

 31 (37)

 H

 EL84-L

 NC

 0

 11 (10)

 H

 H

 NC

 305

 NC

 307

 EL84-L

 NC

 0

 10.5 (10)

 H

 H

 NC

 305

 NC

 307

 ECC83-R

 213 (215)

 19

 33 (37)

 H

 H

 217

 19

 33 (37)

 H

 EL84-R

 NC

 0

 9.7 (10)

 H

 H

 NC

 304

 NC

 306

 EL84-R

 NC

 0

 9.2 (10)

 H

 H

 NC

 304

 NC

 306

 I was puzzled by the discrepancy in the common ECC83 voltages so I looked at R20 and R19. These cannot be read in-situ because of the large electrolytics. I unsoldered them and found R20 was 5.1Kohm and R19 was a whopping 245Kohm. I fitted a 4.3Kohm and a 43Kohm in their place and sure enough the ECC83 anode voltages were a lot higher.

Checking the current there was 60.8 volts across 4.3Kohm which works out at 1.4mA or about 0.7mA for each anode. Across the 4.3Kohm I measured 19.5 volts or 4.5mA which means that the left channel driver is sinking about 1.5mA per anode. The total HT current is now 148mA.

Each of the EL84s is drawing from 34mA (=10.37 watts) to 41mA (=12.4 watts) although this includes the screen grid.

I'll need to commission one of the Pre-Amps to see the final output power.

 

 VALVE

 PIN 1

 PIN 2

 PIN 3

PIN 4

PIN 5

PIN 6

PIN 7

PIN 8

PIN 9

 ECC83-C

 185 (135/160)

 0

 1.8

 H

 H

 187 (135/160)

 0

1.8 

 H

 ECC83-L

 210 (215)

 20

 39 (37)

 H

 H

 208 (210)

 20

 39 (37)

 H

 EL84-L

 NC

 0

 11 (10)

 H

 H

 NC

 305

 NC

 307

 EL84-L

 NC

 0

 10.5 (10)

 H

 H

 NC

 305

 NC

 307

 Having analysed the condition of the parts and got it working one might wonder what sort of audio output power is this amplifier capable? Well, it was designed many years ago when output figures were true and not faked. The key parameter is distortion and the Leak Stereo 20 can deliver (per channel) 10 watts RMS at 0.1% distortion at 1000Hz. Frequency response is from 20Hz to 20KHz. Bump up the input and one can get 11 watts at the expense of some distortion.

 

 Now, I'll tackle the first of the two "Leak Point One Stereo" pre-amplifiers carrying the stamp "Lund Brothers October 1958"

 

 Below are views of the pre-amplifier

Above you can see the provision for adjusting three of the inputs to make them match in volume, presumably to match the volume for the Tape input which doesn't have a trimmer pot. The octal plug connects via a cable to the Leak amplifier. This connection carries audio which is slightly odd because one would have expected a twin screened phono lead to have been specified. Looking at the markings above the phono outputs are marked "RECORD" and these are fixed and do not include the volume control. I guess these could be used to connect to the amplifier phono inputs, although the octal connections with their drawbacks are still in place.

Below.. the top view showing the four EF86 valves. The preamp appears to have been modified long ago when the volume control twin gang potentiometer was replaced. The original had a single gang switch which was wired to those twisted yellow wires. That feature was designed to be used to switch the mains supply in the associated power amplifier, however in both the Leak amplifiers that arrived with the preamp that particular feature wasn't wired up.

  The black cylindical Hunts capacitors all leaked around 40uA (= about 750Kohm) with my 31 volt power supply connected.

These capacitors are oil/paper types no better than others of the same composition.

The yellow Plessey electrolytics mostly measured 75uF.

I fitted fourteen 100nF x 500V chips and four 47uF x 63V electrolytics. that green capacitor in the picture below was marked 8uF and measured 10uF with zero ohms ESR so I'll leave it in place.

 

The view below shows the circuit board carrying replacement capacitors.

 

I checked the resistors and found most were high but within 20% except the 10K which were up to 15K.

I decided to power up the preamp and compare the voltages against those in the circuit diagram below and for reasonable balance between the left and right circuits. If there's any serious imbalance I'll check and replace resistors if necessary.

 

Click to see the circuit diagram full size

 Above is the test bench for the amplifier. The aim was to identify any differences between the pre-amp output channels. In a quiescent state, with no input and set to the Tuner input circuit, I switched on.

After a warm-up period of a few minutes I checked the voltages at the four EF86 valve bases. The starred headings carry the voltages taken from the circuit diagram shown above.

 VALVE

 ANODE

 ANODE*

 SCREEN

 SCREEN*

 CATHODE

 CATHODE*

 V1R

 69.6V

 65V

 40.8V

 40V

 1.25V

 -

 V1L

 69.6V

 65V

 40.8V

 40V

 1.15V

 -

 V2R

 90.9V

 80V

 41.2V

 40V

 1.17V

 1.3V

 V2L

 92.7V

 80V

 32.2V

 40V

 1.04V

 1.3V

I decided to check the resistors around V2R and V2L and swapped four 2.2Mohm, two 220Kohm and two 10Kohm at anodes, screens and control grids. This fixed the imbalance between the output valves and brought the voltages closer to those in the circuit diagram.

Next I fed audio signals into the Tuner input to confirm the left and right output channels were working. I used a small loudspeaker on the left channel and a dummy load on the right.

 Starting with a tone of 1KHz at an input voltage of 6.1mV with a 40dB attenuator the outputs measured 1VRMS and 1.3VRMS. The speaker was 8 ohms and the dummy load (shown here) is 9 ohms.

I adjusted the balance control and set the outputs about equal then tried different frequencies. The outputs remained fairly flat and balanced from 20Hz to 10KHz. Next I used 10KHz (as this is way outside my hearing) and found the maximum output (again measured on my multimeter) was about 7.4 volts RMS at the dummy load. This represents a power output of....

(7.4 x 7.4)/9 = 6.08 watts or about 12 watts for the two channels.

These figures are about half that in the specification so my next test will use my oscilloscope.With my 9 ohm dummy load I should be able to see 10 volts RMS before distortion is evident. I can also check the other pre-amp inputs are OK.

 The next day I had really challenging problems when I used my oscilloscope. The first issue was bad leads. A bad twin phono lead, a bad croc clip lead and a bad scope lead all puzzled me until I checked these with my multimeter. Once sorted out, I looked at the signals to the grids of the EL84s and noticed a ripple which was superimposed on the test sinewave. By trial and error I found this could be much reduced by decoupling the two HT feeds to the ECC83s and the EF86s. I then found the outputs at the speaker terminals were lower than I'd read with my multimeter (8 volts and 5 volts). Further checks will be needed.. in particular I need to check the main HT line for voltage and ripple.

I have a lot of high capacitance, high voltage capacitors from old lift equipment but these are unsuitable for valve rectifier reservoir use. Howevr I added one at the smoothing side of R21 and a second at the end of R20. These reduced the 100Hz ripple at the smoothed HT from 3600mV to 360mV and at the EL84 grid capacitors to 34mV. The monitored sinewave test inputs now show on the scope as steady sinewaves rather than with heavy superimposed 100Hz ripple. The HT is 318 volts rather than the 325 volts shown in the schematic. Maximum undistorted output is still low however at 8 volts into 9 ohms or 7 watts RMS. The output should be 9 volts.

Further testing revealed something interesting. I'd used the balance control on the preamp to set the speaker outputs about the same level but I then discovered the preammp outputs were miles different so I set the input voltages at the grids of the first ECC83 to about the same level of 75mV. I then looked at the sinewaves along the main amplifier. Prior to that I checked the cathode voltages of the EL84s . The Left channel were 10 volts and the Right channel 9 volts with no audio input. Turning on audio made the Left channel cathodes rise to circa 15 volts but the Right channel remained at circa 9 volts. I traced the discrepancy to the anodes of the ECC83 which measured 11.2 volts RMS and 6.6 volts RMS. This difference was echoed at the cathodes of the following ECC83s which were 226mV and 123mV. The obvious thing was to swap the common ECC83 but this made no change and neither did swapping the ECC83 on the Right channel.

I'd forgotten that when I'd first checked the amplifier the speaker output setting was 16 ohms on both channels which is unusual as most loudspeakers are either 4 ohms or 8 ohms. I changed this to 8 ohms and also snipped the feedback resistors R18 and the cathode decouplers C1. This changed things completely and I was able to crank up the Right channel to above the rated figure of 11 volts but this time the Left channel was behind. I hink this is probably because I'm using an old loudspeaker. I'll now replace this with a dummy load. In fact I'll add a parallel 72 ohm resistor to the other and make this 8 ohms instead of 9 ohms.

The final testgave me a pair of equal amplitude sinewaves at 1KHz with an amplitude of about 9 volts which represents between 10 to 11 watts but with negative feedback components disconnected. I'll test these and fit new parts as necessary.

 I tested the pair of cathode capacitors at the first ECC83 and found they leaked so replaced these with new 1nF capacitors. surprisingly the amplifier worked more cleanly producing what look like perfect 9 volt RMS 1KHz sinewaves into the dummy loads which measured 7.5 ohms and 7.8 ohms. I adjusted the balance control to get the output shown.

 

The power into the loads is now:-

(9.44 x 9.44)/7.8 = 11.42 watts

(9.23 x 9.23)/7.5 = 11.36 watts

 

These figures are exactly as the manufacturer specified.

 

 

 The picture isn't very clear but the signal generator is producing 1.6 volts RMS fed in parallel to the Tuner input phonos. The amplitude control and the pre-amp volume control together set the levels shown above.

 

Below, final views of the underside of the amplifier and pre-amp, the former now fitted with additional HT smoothing capacitors.

Note the toggle switch fitted in place of the shorting link at the unused remote power switching point.

 

 The table below shows the results of final tests. I didn't adjust the preset pots on the preamp so these are still set up for the last owners hi-fi setup. The signal generator was connected to each pair of phono sockets in turn and the generator output set to about 1KHz and increased from zero to maximum undistorted output into 8 ohm dummy loads. The left and right outputs were set to the same level by tweaking the balance control.

The output voltages indicated on the oscilloscope were about 9.3 to 9.5 volts RMS equating to the makers maximum undistorted output of 11 watts per channel. The preamp output of 125mV is said by the maker to produce the rated power for the two channels. During the tests the preamp volume control was set to about half way hence the differences between measured sensitivities and those specified.

 INPUT

 VOLTAGE FOR 9 VOLTS RMS OUTPUT

MEASURED

 VOLTAGE FOR 125mV FROM PREAMP

FROM SPECIFICATION

 TAPE

 41mV

 3mV

 TUNER

 1.45V

 35mV

 INPUT PICKUP

 16mV

 3.5mV

 EXTRA

 1.5V

 35mV

 MIC

 9mV

 2mV

Now for the second Leak Stereo 20 Amplifier

 

 

 The condition of the components in the second example was worse than the first. In fact the mains fuse had reportedly blown leading the owner to suspect the amplifier was suffering from a bad transformer.

I checked the DC resistances of the mains transformer and the pair of output transformers and found nothing amiss, however state of numerous components certainly implicates a problem with the output valves.

Then a further quick check revealed that the two parts of the condenser (C10 & C11 32uF above- top right) were open circuit and it's possible a short here (in conjunction with excessive output valve anode currents) blew the fuse and opened the connections inside the can.

As with the first amplifier the mechanical designer (or maybe the electronics designer or drawing office??) has mounted the components in a nicely symmetrical pattern ignorant of the niceties of condenser markings. The large 0.25uF coupling condensers are marked with a black band which normally indicates the outer foil connection. Ideally the outer connection should be used for the circuitry prone to least sensitivity not the grid circuit with its high impedance. The way they're connected might result in a little hum on one channel.

If you compare this picture with that for the first amplifier you'll see there are two condensers missing. These will have been identical to the pair of green ones on the left and would have been baked by the adjacent cathode resistors. Those read 7.1 Kohm and 5.7 Kohm instead of a mere 270 ohms and have burnt up due to leakage in a pair of those 0.25uF Metalpack condensers each driving their associated output valve anode current too high. Oddly the mates of these must have been less leaky as the 270 ohm pair haven't degraded as much as the others although they are still out of spec, being 336 ohms and 429 ohms. Another resistor in poor shape is the 47 Kohm feeding the first amplifier valve which is a whopping 31 times its marked value at 1.48 Mohm.

Finally there's a 10K at 18K and a pair of 22 Kohm resistors at the first amplifier reading 28K and 32K. This is really funny because the owner was sold an ECC83 with matching triodes. The ECC83 anode resistors are marked 100K and not too bad although the circuit diagram shows these were originally (or should be) 47 Kohm.

The first task was to replace all the old condensers then to fit several new resistors as shown below.

Next I need to add a couple of high voltage electrolytics to smooth the HT supply as I did with the first amplifier.

 

 Almost the same collection of bad components as the first amplifier.

The four large Metalpack condensers marked 0.25uF (=200nF to 300nF) measured 316nF, 347nF, 410nF and 504nF.

The pair of Visconol 0.1uF (=100nF) were 442nF and 482nF.

The two Plessey electrolytics marked 50uF were 112uF and 238uF.

The best examples were that pair of 0.02uF Metalmite condensers which fooled my Chinese tester to be resistors of 221 Kohm and 547 Kohm.

Of course it isn't so much the capacitance change that matters it's the leakage at HT voltages. Large Metalpack leakages were responsible for those four burnt 270 ohm cathode resistors.

I'm leaving the HT smoothing condensers physically in place as they form part of the look of the amplifier.

Finally I fitted a On/Off toggle switch in place of the shorting link for the remote power switch feature.

 

 Now for the second "Leak Point One Stereo" pre-amplifier.

 

 All the resistors tested within an acceptable range, but of course it was a different matter when it came to the old condensers as these were all leaky and in some cases cracked open. I fitted fourteen 100nF chip capacitors rated at 500VDC and four 100uF x 16V electrolytics. I used a 33uF electrolytic in place of the old 8uF HT smoothing condenser. I also fitted a pair of new 6.8nF in place of 10nF TCC condensers and a pair of 47nF in place of two cracked open Hunts condensers.

 

 Now I'll test the pre-amp with the main amplifier. I tested the latter and found the HT rectifier (GZ34) lit up with a bright purple glow accompanied by crackling. This explained the fact that the amplifier had been set aside when its fuse blew many years ago. At the time a bad transformer had been suspected which would have made a repair extremely expensive. Thankfully this was not the case (just a gassy rectifier) and a replacement GZ33G rectifier brought the amplifier back to life, however, when I checked the data on the GZ32 I found the heater was rated at 2.3A compared with 1.9A for the GZ34 and checking the heater voltage showed it was 4.7 volts with the GZ32 in place. Another option is to disconnect the GZ34 heater and fit a pair of silicon diode rectifiers instead of a new valve. That solution would increase the HT voltage and raise the audio power output. I checked the HT and found it was 327 volts at the reservoir and 310 volts at R21 meaning the HT current (withut the preamp) is 170mA. However, raising the HT from its 310 volts to something like 350 volts would increase the anode dissipation, and reduce the life, of the EL84s which are rated at up to 300 volts. If this change was made I'd increase the value of R21 from 100 ohm to something like 200 ohms.

 

The second amplifier + preamp were missing a connecting cable and lately I'd identified lots of Scart cables now redundant. These old cables are ideal for use as interconnecting cables as they use several screened wires. I tried using one but in the end it was too fragile so reverted to just ordinary wires.

OCTAL PIN

 PURPOSE

1

 GROUND

2

 RIGHT AUDIO OUTPUT

3

 HT 135 VOLTS

4

 HT 195 VOLTS

5

 LT 6.3 VOLTS AC

6

 NOT USED

7

 NOT USED

8

 LEFT AUDIO OUTPUT
 
 
 
 

 Waiting for repairs

 

Return to Reception