Hi Friends I bought a pair of quad esl 57 speakers. The speakers weren't in use for about four years. After I power them up, i let them warm about one day, then I tried to listen some music. The result wasn't so good, i felt like the speakers need a lot of power to play music! I decided to open the back and take some measurements. The input voltage from the transformer to the power supply board was 635V. The High voltage for the treble was 1,5KV and for the Bass 3,69KV I tested both with panels and without the panels, was the same. The other speaker has almost the same measurements.
I decided to order some new resistors, capacitors and diodes and make a new power supply. After received all materials, I made one new power supply and I was surprised from the new measurements! The High voltage for the treble was 1,02KV and for the Bass 1,87KV. The Parts I used are: Resistor 330K 0,5W metal film 1% Resistor 2M2 2W metal film 1% Capacitor 0,01uF 3KV 50pcs High Voltage Ceramic Disc Capacitors 10000PF 10NF 0.01uf 3000V 3KV 103 NEW | eBay Diode R3000 100pcs NEW R3000 DIODE High VOLTAGE RECTIFIER 3000V 1A | eBay
Does anyone has idea what is wrong?
I decided to order some new resistors, capacitors and diodes and make a new power supply. After received all materials, I made one new power supply and I was surprised from the new measurements! The High voltage for the treble was 1,02KV and for the Bass 1,87KV. The Parts I used are: Resistor 330K 0,5W metal film 1% Resistor 2M2 2W metal film 1% Capacitor 0,01uF 3KV 50pcs High Voltage Ceramic Disc Capacitors 10000PF 10NF 0.01uf 3000V 3KV 103 NEW | eBay Diode R3000 100pcs NEW R3000 DIODE High VOLTAGE RECTIFIER 3000V 1A | eBay
Does anyone has idea what is wrong?
I think your original EHT block is not faulty. The mid/treble voltage is about OK. For measuring the bass panel bias voltage, you need an instrument having at least 100 Mohm input resistance, or even higher. The sensitivity of the ESL is 10 dB below the common cone speakers. If the frequency balance is good, they are probably OK.
Hi Klarskov
I double check all connections, all looks fine to me.
I attach the schematic, if you see something wrong please let me know.
Hi lcsaszar
My multimeter have internal resistance 10Mohm and I am using a HV probe of 1000Mohm. I can't test the bass panel bias voltage this way?
I thought that Quad ESL57 have 93dB sensitivity. Isn't correct?
I am using an transistor 100W/8ohm amplifier and usually with the volume at 10 to 11 o'clock I make a party, but with quad's is quiet and I afraid to turn the volume more
I'm not sure your terminal blocks can' candle 4500 volts between terminals, so you might be loading your supply down with breakdown across the terminal strip.
You are wise to not turn the volume up. The quads should be about as efficient as a cone and dome speaker, not exactly but not hugely different either.
Sheldon
You are wise to not turn the volume up. The quads should be about as efficient as a cone and dome speaker, not exactly but not hugely different either.
Sheldon
Hi Sheldon
You think Is better to remove the terminal block and leave only pin wires to solder?
I put all parts too close, maybe I'll make it from the beginning leaving more space from each part
The fact that the woofer voltage at the end of 8 multiplier sections is drooping more than the tweeter voltage at the end of 2 sections makes me think it is either leakage problems, as stokessd already mentioned, or the capacitors have a smaller value than marked. Can you measure and confirm capacitors are indeed 10nF and not something smaller like 1nF?
You might also measure the voltage across the 2.2M resistor to confirm it is as expected.
I use a 1500 meg resistor based HV probe with my fluke 87 (10 meg input impedance). A properly functioning EHT block will read about 10.5v on the treble tap and about 32v on the bass section. The ability to source current drops with each stage of a cockroft walton multiplier and you see that in my measurements. The bass reading should be 4X the treble tap and it's just over 3X. Your first measurement actually look like your original EHT modules were ok. Keep in mind that the quad power supplies are REALLY wimpy the 1gig probe will pull them down a lot.
If you have EHT problems, both will rarely (and I mean rarely) be the same. so your stereo image will be all lopsided. The fact that both speakers were measuring the same sounds suspicious, like any issue you have isn't in the power supplies. Maybe let them charge longer. Old panels can take a LONG time to charge because that crappy conductive coating evaporates over time.
sorry to not have any better info for you,
Sheldon
If you have EHT problems, both will rarely (and I mean rarely) be the same. so your stereo image will be all lopsided. The fact that both speakers were measuring the same sounds suspicious, like any issue you have isn't in the power supplies. Maybe let them charge longer. Old panels can take a LONG time to charge because that crappy conductive coating evaporates over time.
sorry to not have any better info for you,
Sheldon
Another possibility ---- following the PASS forum on diyaudio, there are lots of warnings in that forum to not buy counterfeit semiconductors on EBay (especially rare JFETS and especially from China/Hongkong). No proof that your diodes are bad, but the price USD 2.61 for 100 pieces seems a bit too good to be true.
I had problems (in Europe) to source the diodes used by Quad, GP02-30 with decent delivery times , but found these by Diotec (German company):
5000 V and 3 uA leakage current (lower that the QUAD parts)
https://www.elfa.se/Web/Downloads/r_/ds/hv3_hv5_eng_ger_ds.pdf
Price around USD 3 for 10.
They work well in the ESL57 circuit (got 1.55, 6.3 kV)
Alternatively you can of course use two plain 1N4007 in series.
Just a theory, might also be cabling and/or connectors as mentioned above...these circuits are very wimpy (in my case I found that even 55 years old dirt on cables make them lose voltage, and cause hum in bass panels.
...and with the same suspicious mind ...check the capacitors too (as already mentioned)
I had problems (in Europe) to source the diodes used by Quad, GP02-30 with decent delivery times , but found these by Diotec (German company):
5000 V and 3 uA leakage current (lower that the QUAD parts)
https://www.elfa.se/Web/Downloads/r_/ds/hv3_hv5_eng_ger_ds.pdf
Price around USD 3 for 10.
They work well in the ESL57 circuit (got 1.55, 6.3 kV)
Alternatively you can of course use two plain 1N4007 in series.
Just a theory, might also be cabling and/or connectors as mentioned above...these circuits are very wimpy (in my case I found that even 55 years old dirt on cables make them lose voltage, and cause hum in bass panels.
...and with the same suspicious mind ...check the capacitors too (as already mentioned)
Last edited:
Hi,
possible ´faults´ and workarounds:
- input connector with too small RM .... use at least RM7.5.
- 10nF Caps are quite small ... I´d rather choose >47nF, especially with higher numbers of cascade sections
- take care that solder joints are rather rounded (like a blob not a peak)
- throroughly clean the PCB and coat it with some isolating spray (Plastik70 or alike)
- routing isolating slots between points of highly differing potentials ensures that no leakage path develops over time, be it a sweaty fingerprint, or dust in combination with air humidity
jauu
Calvin
possible ´faults´ and workarounds:
- input connector with too small RM .... use at least RM7.5.
- 10nF Caps are quite small ... I´d rather choose >47nF, especially with higher numbers of cascade sections
- take care that solder joints are rather rounded (like a blob not a peak)
- throroughly clean the PCB and coat it with some isolating spray (Plastik70 or alike)
- routing isolating slots between points of highly differing potentials ensures that no leakage path develops over time, be it a sweaty fingerprint, or dust in combination with air humidity
jauu
Calvin
I tested all components, capacitors are 10 to 12nF, diodes have 1,5V voltage drop. I also made a simple half wave power supply with each diode, input 20VAC and the output was 28VDC.
The input voltage from the transformer is 635VAC and after the voltage divider is 550VAC
Last edited:
Hi sheldon
I left them 3 days after my first listening test, the result was the same..
Maybe you have right and the original power supply is ok
I think is better to move one speaker to a friend who has also quad esl57 and compare..
I couldn't find GP02-30 that's why I bought R3000. I never had problems with diodes and rectifiers from china.
Thanks for sharing your experience. I'll buy some diotec HV5 to replace R3000
How did you measure the high voltage? with a HV probe? and what is the resistance of the probe ?
On the topic of measuring the EHT voltage:
stokessd wrote:
This is very similar to how I measure as well, using a Metrix Mtx multimeter (10 meg input impedance) and built a 1000 meg and 2000 meg HV probe, using 3x1Gohm resistors, some flexible plastic tube used for electrical wiring in buildings filled with beeswax (see pic). Measuring with both probes means it is possible to calculate the true voltage without load as well as the generator impedance of the EHT (in the order of 250 meg if the EHT is OK).
stokessd wrote:
This is very similar to how I measure as well, using a Metrix Mtx multimeter (10 meg input impedance) and built a 1000 meg and 2000 meg HV probe, using 3x1Gohm resistors, some flexible plastic tube used for electrical wiring in buildings filled with beeswax (see pic). Measuring with both probes means it is possible to calculate the true voltage without load as well as the generator impedance of the EHT (in the order of 250 meg if the EHT is OK).
Attachments
Last edited:
Today I made the following simple test.
I used a variac to supply the original power supply board with 10VAC, the voltage after the divider was 8,7VAC
Measuring with my multimeter (10mohm internal resistance) the output voltage was 12,55VDC (at 1,5KV tap) and 15,7VDC (at 6KV tap)
Measuring with my multimeter (10mohm internal resistance)+ HV probe(1Gohm) the output voltage was 23,1VDC (at 1,5KV tap) and 76,2VDC (at 6KV tap)
The conclusion is that probably I am shorting the output..
The question is what should be the minimum impedance to test the 6kV tap?
The other question is, if the power supply is fine, what I should check?
I used a variac to supply the original power supply board with 10VAC, the voltage after the divider was 8,7VAC
Measuring with my multimeter (10mohm internal resistance) the output voltage was 12,55VDC (at 1,5KV tap) and 15,7VDC (at 6KV tap)
Measuring with my multimeter (10mohm internal resistance)+ HV probe(1Gohm) the output voltage was 23,1VDC (at 1,5KV tap) and 76,2VDC (at 6KV tap)
The conclusion is that probably I am shorting the output..
The question is what should be the minimum impedance to test the 6kV tap?
The other question is, if the power supply is fine, what I should check?
Last edited:
I used a single 1 Gohm resistor for the 1000 meg short probe and
2 x 1 Gohm in series for the longer 2000 meg probe.
I would say any >1G probe should work well.
Taking your 10 VAC in and the 76,2 VDC out and scaling it up linearly to the 635 VAC you measured in the 57 circuit you get 4339 V... furthermore assuming that 250 Mohm is the typical EHT impedance you get the real voltage (with your 1000 meg probe) as 4339*(1000+250)/1000 = 6048 V (good). Not sure of what conclusions to draw with certainty, but you seem to have the correct Cockroft-Walton characteristics at low voltages (your circuit is 100% correct and your diodes and caps work as intended at low voltages, but you get leakage currents bogging everything down as voltage increases => follow Calvin's advice list and continue to suspect caps or diodes HV behaviour.
/my 5 cents
2 x 1 Gohm in series for the longer 2000 meg probe.
I would say any >1G probe should work well.
Taking your 10 VAC in and the 76,2 VDC out and scaling it up linearly to the 635 VAC you measured in the 57 circuit you get 4339 V... furthermore assuming that 250 Mohm is the typical EHT impedance you get the real voltage (with your 1000 meg probe) as 4339*(1000+250)/1000 = 6048 V (good). Not sure of what conclusions to draw with certainty, but you seem to have the correct Cockroft-Walton characteristics at low voltages (your circuit is 100% correct and your diodes and caps work as intended at low voltages, but you get leakage currents bogging everything down as voltage increases => follow Calvin's advice list and continue to suspect caps or diodes HV behaviour.
/my 5 cents
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.