Re: Powersupply?
Both. I didn't use a capacitance multiplier because I had some very big capacitors, and because I didn't want the voltage drop of the multiplier. But normally I would use it, because it makes good engineering sense to use one, and there have been good reports about using it. And you still need a large smoothing capacitor before the multiplier.
roger-k said:
Both. I didn't use a capacitance multiplier because I had some very big capacitors, and because I didn't want the voltage drop of the multiplier. But normally I would use it, because it makes good engineering sense to use one, and there have been good reports about using it. And you still need a large smoothing capacitor before the multiplier.
capacitance multiplier
I am using a capacitance multiplier with good results and a 45000uF (3x15000) per rail before it. The only problem is the voltage drop if you would like to be sure if the voltage of the mains drops arround 10% at maximum so the minimal voltage drop should be around 7-8V that means that the pass transistor will dissipate around 15-16W....
Pred
I am using a capacitance multiplier with good results and a 45000uF (3x15000) per rail before it. The only problem is the voltage drop if you would like to be sure if the voltage of the mains drops arround 10% at maximum so the minimal voltage drop should be around 7-8V that means that the pass transistor will dissipate around 15-16W....
Pred
OK. Thank for the answers.
15000uF.... This is more than enough?
Rod Elliott use 2x4700uF:
http://sound.westhost.com/project15.htm
15000uF.... This is more than enough?
Rod Elliott use 2x4700uF:
http://sound.westhost.com/project15.htm
roger-k said:
The schematic showing 4700uF was prepared to suit a current draw of 1.25A. 4700uF is the minimum requirement, as Rod indicates further on in the text:
"Increasing the capacitance (especially at the input) is recommended, and I would suggest 10,000uF as the ideal instead of the 4700uF shown."
Fire works
the last few days I've been making some fireworks by blowing fuses and transistors in my attempt to get my JLH running (as per JLH-update from Geoff's great site)
After checking and rechecking and rerechecking I ask help.
I've a nice cap-multiplier power supply running (rails + and - 26 V).
I've soldered a jlh update with 2n3906 input; 2n3610 driver; TIP3055 output (later BC245c), dc offset circuitry with two bc560 and the Iq control with BC560 and BD140.
When I power the thing up I've just about time to do some measurements before producing smoke.
I've a voltage drop of about 0.5V over the 0R33 which means an Iq of about 1,5 A.
I've a huge DC-offset of about 26V. And then the fuse and/ or output transistors blow.
without the ouput transisors connected I'v a Dc offset of about 3V.
Is there a way of testing my circuit or parts of it without producing smoke? What can I do to find the fault (visual inspection only confirms that I've soldered everything correct.) I did test al the transistors of course. They are fine.
Help, I cannot find the problem
the last few days I've been making some fireworks by blowing fuses and transistors in my attempt to get my JLH running (as per JLH-update from Geoff's great site)
After checking and rechecking and rerechecking I ask help.
I've a nice cap-multiplier power supply running (rails + and - 26 V).
I've soldered a jlh update with 2n3906 input; 2n3610 driver; TIP3055 output (later BC245c), dc offset circuitry with two bc560 and the Iq control with BC560 and BD140.
When I power the thing up I've just about time to do some measurements before producing smoke.
I've a voltage drop of about 0.5V over the 0R33 which means an Iq of about 1,5 A.
I've a huge DC-offset of about 26V. And then the fuse and/ or output transistors blow.
without the ouput transisors connected I'v a Dc offset of about 3V.
Is there a way of testing my circuit or parts of it without producing smoke? What can I do to find the fault (visual inspection only confirms that I've soldered everything correct.) I did test al the transistors of course. They are fine.
Help, I cannot find the problem
Let try to help
I will try to help you debug your problem at least a little bit.
First what you should do is to adjust with the potentiometer that your Iq is as small as possible. So if I remember well you should put it at highest resistance. In that way your Iq will be much less than an amp and you can slowly debug/measure your circuit. I would first put the pot 3 times the value of the proposed one and reduce the current of the output to 100-200mA in order to debug and than slowly rise up to the full current value. Wait couple of hours between the measurements to see if the Iq/offset is stable.
I do not now (or not remember the details) of the design, it is 1 year since I built mine. What I suspect is the bigest problem is your output transistor. If you use TIP3055 their dissipation is maximum 90W at 25deg C and are aroun 50-60 at 75W. Are you using two in parallel? If not, I think you are going into the thermal runaway and blow the devices. If I remember well you should use TO3 2N3055 devices, they have 20W more capability of dissipation and less thermal resistance. Did you use heavy heatsink and follow good thermall procedures. You should remember that you are, without inpit signal dissipating 26x1.5=40W per device. Are your drivers on their own heatsinks? They dissipate as well and if not stable could drive the output transistors to consume more and more current. I won't judge your choices, but for me your rails are high. I made JLH with MJ15003 18V rails and 2A of Iq current and with enormous heatsinks still temperature goes to 45-50 degrees on the haetsinks and the dies of the transistor is much higher. In my setup I also made couple of mistakes (turned driver transistor upside down) but if your thermal management is ok than there is a little chance to blow the things.
Good luck and keep us on your progress. Wanderfull amp ( I call it last of solid state for me) and it will pay of good sound when you finish.
Pred
I will try to help you debug your problem at least a little bit.
First what you should do is to adjust with the potentiometer that your Iq is as small as possible. So if I remember well you should put it at highest resistance. In that way your Iq will be much less than an amp and you can slowly debug/measure your circuit. I would first put the pot 3 times the value of the proposed one and reduce the current of the output to 100-200mA in order to debug and than slowly rise up to the full current value. Wait couple of hours between the measurements to see if the Iq/offset is stable.
I do not now (or not remember the details) of the design, it is 1 year since I built mine. What I suspect is the bigest problem is your output transistor. If you use TIP3055 their dissipation is maximum 90W at 25deg C and are aroun 50-60 at 75W. Are you using two in parallel? If not, I think you are going into the thermal runaway and blow the devices. If I remember well you should use TO3 2N3055 devices, they have 20W more capability of dissipation and less thermal resistance. Did you use heavy heatsink and follow good thermall procedures. You should remember that you are, without inpit signal dissipating 26x1.5=40W per device. Are your drivers on their own heatsinks? They dissipate as well and if not stable could drive the output transistors to consume more and more current. I won't judge your choices, but for me your rails are high. I made JLH with MJ15003 18V rails and 2A of Iq current and with enormous heatsinks still temperature goes to 45-50 degrees on the haetsinks and the dies of the transistor is much higher. In my setup I also made couple of mistakes (turned driver transistor upside down) but if your thermal management is ok than there is a little chance to blow the things.
Good luck and keep us on your progress. Wanderfull amp ( I call it last of solid state for me) and it will pay of good sound when you finish.
Pred
I posted a recommended startup procedure a while ago:
http://www.diyaudio.com/forums/showthread.php?postid=1067918#post1067918
http://www.diyaudio.com/forums/showthread.php?postid=1067918#post1067918
debugging the fireworks continued
Thank you for your help, Gives me courage again...
I definately have no thermal problems. The devices blow before they even get warm.
Well, I had the transformers handy, so I've used them. I expect the rails to drop to 23 or 24 V when it is feeding 2 channels on the right Iq.
I've two LARGE heatsinks: size: about 30cm X 60 cm, fin length 8 cm, total weight 10 kg a piece. That should keep things cool, or am I mistaken
?
I plan to put 2sc5200 in parallel when everyting runs ok. Then I'll adjust the Iq to obtain a reasonable operating temp.
thank you all...
Thank you for your help, Gives me courage again...
Well, my pot doesn't allow a lower IQ than 1,5 A. I will put a series resistor in and use the suggested 100 to 200 mA.
I definately have no thermal problems. The devices blow before they even get warm.
Well, I had the transformers handy, so I've used them. I expect the rails to drop to 23 or 24 V when it is feeding 2 channels on the right Iq.
I've two LARGE heatsinks: size: about 30cm X 60 cm, fin length 8 cm, total weight 10 kg a piece. That should keep things cool, or am I mistaken
? I plan to put 2sc5200 in parallel when everyting runs ok. Then I'll adjust the Iq to obtain a reasonable operating temp.
Yes, I thought so too and was true for the 1969 (original schematics) version I've build previuosly.
What is the i/p transistor? the input transistor?
thank you all...
Re: debugging the fireworks continued
) I put the schemativcs in LT Spice and simulated the circuit. Then I had all DC signals at critical points. When I comapred them with the measurements it was fairly close. In that way I could debug the circuit quite easy. In your place I would try to calm Iq first as I suggested and then go easy to see where further problems are...
Pred
This is quite strange ynd you should keep a close look at your PCB and components. As you can read in previous posts you should have around 600mA for the maximum 50ohm value of the Iq pot (If you use the parallel devices). Oscillation could be a problem but I am not sure you could read it as 1.5Amps of Iq or the offset at the output as high as you have. But worth of checking. What I did at the time (unfortunately all is lost by my hard crach this summerdeduikertjes said:
) I put the schemativcs in LT Spice and simulated the circuit. Then I had all DC signals at critical points. When I comapred them with the measurements it was fairly close. In that way I could debug the circuit quite easy. In your place I would try to calm Iq first as I suggested and then go easy to see where further problems are...Pred
Problems solved
Well, tanx to all your help the problems are solved.
First I connected the amp to a lower voltage PS (10 V rails), and added series resistance before the Iq pot.
Then I could measure without smoke. In the end it proved to be the input transistor 2n3906. I've replaced it with bc550c and after that I was winning.
I could increase Iq and get the DC offset to 0. The amp plays some music.
On Geoff's class-A site the DC values for all leads of all transistors are given. Now my measurements are really close
Thanks, MArco
Well, tanx to all your help the problems are solved.
First I connected the amp to a lower voltage PS (10 V rails), and added series resistance before the Iq pot.
Then I could measure without smoke. In the end it proved to be the input transistor 2n3906. I've replaced it with bc550c and after that I was winning.
I could increase Iq and get the DC offset to 0. The amp plays some music.
On Geoff's class-A site the DC values for all leads of all transistors are given. Now my measurements are really close
Thanks, MArco
Congratulations! Enjoy!
I've placed all the new information I received from Geoff (schematics, layouts, gerbers, etc.) on my personal webspace at
http://members.shaw.ca/paul.r.brown/DIY/Geoff Moss JLH/
This is the same stuff I posted individually here a few pages ago. Geoff is still planning to update his site but hasn't got to it yet.
I've placed all the new information I received from Geoff (schematics, layouts, gerbers, etc.) on my personal webspace at
http://members.shaw.ca/paul.r.brown/DIY/Geoff Moss JLH/
This is the same stuff I posted individually here a few pages ago. Geoff is still planning to update his site but hasn't got to it yet.
Paul, thanks for the link. I am however confused by a couple of things.
The version of the schematic is of the high power option only.
I thought that there was an extra capacitor and an RV3 on Geoff's later revision, plus removal of the feedback cap. Some of the transistors differ too.
I built the 1996 jlh several years ago but am about to build a new one including the latest revisions. I have to say that having trawled through most of the posts on this thread I'd love a summary like the one you've done for the lower power version too.
Speaking of the els57 version, would that one be recommended for use with els63s too? I used to use my jlh1996 with my '63s.
The version of the schematic is of the high power option only.
I thought that there was an extra capacitor and an RV3 on Geoff's later revision, plus removal of the feedback cap. Some of the transistors differ too.
I built the 1996 jlh several years ago but am about to build a new one including the latest revisions. I have to say that having trawled through most of the posts on this thread I'd love a summary like the one you've done for the lower power version too.
Speaking of the els57 version, would that one be recommended for use with els63s too? I used to use my jlh1996 with my '63s.
United
The pcb that Paul has linked to is a 'universal' layout that will accommodate both the standard and higher power versions from my 'update' page. It will cater for various capacitor sizes and alternative transistor leadouts along with some optional features such as the output Zobel network and larger heatsinks for Q3 and Q8.
The RV3 arrangement for trimming the output dc offset variation between switch-on and normal operating temperature has not been included as feedback from several constructors indicated that the small benefits gained from the far more lengthy setting up procedure were not worth the effort. I would also recommend retaining the dc blocking capacitor in the feedback network, but anyone who wishes to omit it can simply use a link on the pcb.
The higher power version is suitable for 57s and 63s, along with other speakers that present a low impedance load, and also where more power is required than can be obtained from the standard circuit.
I have another, smaller pcb layout for the single output transistor version. This is 100 x 80 so that two boards can be etched on a single eurocard. The board can be single or double sided but space constraints mean that the input capacitor and rail fuses have to be mounted on the chassis and the pads for Q4 are laid out to suit the 2SA970 (a BC560 can be used by carefully crossing the leads). I can supply a high-res image of the copper side for home etching or Gerber files for commercial production.
The transistor changes (Q5/6/7) came about because two or three people had problems with instability in the Q5/Q6 ccs when the 2SA970 was used in this position. This was probably due to the particular layout being used (it occurred before I produced any pcb designs and the amps were built on stripboard or with P-to-P wiring). There were several possible solutions to the problem, for example by adding base stopper resistors, but the easiest cure was to change the transistor type to MPSA56 which I knew from experience worked well in this application even with a less than optimum layout.
Geoff
The pcb that Paul has linked to is a 'universal' layout that will accommodate both the standard and higher power versions from my 'update' page. It will cater for various capacitor sizes and alternative transistor leadouts along with some optional features such as the output Zobel network and larger heatsinks for Q3 and Q8.
The RV3 arrangement for trimming the output dc offset variation between switch-on and normal operating temperature has not been included as feedback from several constructors indicated that the small benefits gained from the far more lengthy setting up procedure were not worth the effort. I would also recommend retaining the dc blocking capacitor in the feedback network, but anyone who wishes to omit it can simply use a link on the pcb.
The higher power version is suitable for 57s and 63s, along with other speakers that present a low impedance load, and also where more power is required than can be obtained from the standard circuit.
I have another, smaller pcb layout for the single output transistor version. This is 100 x 80 so that two boards can be etched on a single eurocard. The board can be single or double sided but space constraints mean that the input capacitor and rail fuses have to be mounted on the chassis and the pads for Q4 are laid out to suit the 2SA970 (a BC560 can be used by carefully crossing the leads). I can supply a high-res image of the copper side for home etching or Gerber files for commercial production.
The transistor changes (Q5/6/7) came about because two or three people had problems with instability in the Q5/Q6 ccs when the 2SA970 was used in this position. This was probably due to the particular layout being used (it occurred before I produced any pcb designs and the amps were built on stripboard or with P-to-P wiring). There were several possible solutions to the problem, for example by adding base stopper resistors, but the easiest cure was to change the transistor type to MPSA56 which I knew from experience worked well in this application even with a less than optimum layout.
Geoff
Yes, each version has slight differences. The parallel output transistors = higher power version is optional.united said:Paul, thanks for the link. I am however confused by a couple of things.
The version of the schematic is of the high power option only.
I thought that there was an extra capacitor and an RV3 on Geoff's later revision, plus removal of the feedback cap. Some of the transistors differ too.
I built the 1996 jlh several years ago but am about to build a new one including the latest revisions. I have to say that having trawled through most of the posts on this thread I'd love a summary like the one you've done for the lower power version too.
Speaking of the els57 version, would that one be recommended for use with els63s too? I used to use my jlh1996 with my '63s.
I'd encourage you to read through the various articles at Geoff's website. It is up to you to decide which changes to pick up. I was hesitant about removal of the feedback cap, for instance, because it sounded too finicky to implement.
The notes from Geoff at my webspace provide some explanation about the version I constructed. Please note that this is Geoff's text, not mine; it is from my correspondence with him. I felt that the information was sufficiently useful to be placed somewhere more accessible than in this thread.
The ESL57 version was designed to provide higher current into a low impedance load. Are your 63s similar impedance? Regardless of the load, I would recommend the parallel output transistors because it makes it easier to transfer the heat from the transistors, so they will run cooler. Or you can set the bias higher.
Hi Geoff, your post arrived as I was composing mine. Thanks for the enlightenment.
I should note that I had no problems whatsoever with oscillation or bias stability (following Geoff's bias adjustment recommendations). The power supply for this amp was much more difficult, actually, as I chose to parallel two power transformer outputs with their primaries wired in series (to use transformers I had on hand).
I should note that I had no problems whatsoever with oscillation or bias stability (following Geoff's bias adjustment recommendations). The power supply for this amp was much more difficult, actually, as I chose to parallel two power transformer outputs with their primaries wired in series (to use transformers I had on hand).