??????
Both the boards were scrupulously cleaned of rosin flux before I installed them, but I'll pull them and reflow the joints.
I just put the funky amp on the bench and unlimbered my trusty DMM (now promised on the Trading Post to someone else).
On turn on, the bias (measured across both emitter resistors), climbed gradually to 27mV and stayed there, and it responded to adjustment predictably. Nothing funny there, with or without soft start. And, on turn on, DC offset showed no bumps, with or without soft start and/or a purely resistive 8 ohm load, in climbing to where I measured it originally at about 6mV. However, when I turned the amp off with no load present the offset climbed off the mV scale and then reversed in polarity and climbed to a - 145mV or so and stayed there until I shorted the outputs.
Anything diagnostic about that behavior?
George😕
Both the boards were scrupulously cleaned of rosin flux before I installed them, but I'll pull them and reflow the joints.
I just put the funky amp on the bench and unlimbered my trusty DMM (now promised on the Trading Post to someone else).
On turn on, the bias (measured across both emitter resistors), climbed gradually to 27mV and stayed there, and it responded to adjustment predictably. Nothing funny there, with or without soft start. And, on turn on, DC offset showed no bumps, with or without soft start and/or a purely resistive 8 ohm load, in climbing to where I measured it originally at about 6mV. However, when I turned the amp off with no load present the offset climbed off the mV scale and then reversed in polarity and climbed to a - 145mV or so and stayed there until I shorted the outputs.
Anything diagnostic about that behavior?
George😕
Hi Andrew,
The dual header strip suggested itself to me on the fly. It was going to be single current and I figured "What the heck?". I'd like to sweep it though.
Hi George,
It looks like it is acting normally from what I can figure. Your trouble should be in the voltage amp stages somewhere, not in the bias area.
-Chris
Same here. Darlingtons are a real pain because the performance spread is so wide.
The dual header strip suggested itself to me on the fly. It was going to be single current and I figured "What the heck?". I'd like to sweep it though.
Hi George,
It looks like it is acting normally from what I can figure. Your trouble should be in the voltage amp stages somewhere, not in the bias area.
-Chris
Colescuttle,
Nothing unusual there. The good thing is that when the amp is on everything seems fine. Are you sure that your pre-amp is working Ok? How about the amplifers input connectors, maybe a bad solder connection there, are they isolated from the chassis? Did you use the audio input terminals on my BOM, or did you solder the inputs directly to the board. If not, I'd check to make sure the terminal screws are nice and tight, and correctly inserted.
Al
Nothing unusual there. The good thing is that when the amp is on everything seems fine. Are you sure that your pre-amp is working Ok? How about the amplifers input connectors, maybe a bad solder connection there, are they isolated from the chassis? Did you use the audio input terminals on my BOM, or did you solder the inputs directly to the board. If not, I'd check to make sure the terminal screws are nice and tight, and correctly inserted.
Al
Colescuttle,
After looking at the pictures you provided, you have a lot going on the input side of your amp, especially with the caps your using, and the audio pot. I'd look very carefully at all your front end connections.
Al
After looking at the pictures you provided, you have a lot going on the input side of your amp, especially with the caps your using, and the audio pot. I'd look very carefully at all your front end connections.
Al
Well, not exactly...
Hi, Al:
The thing that first tipped me off that something was wrong was a big imbalance between channels that had not been there previously. With the passive 20K shunt attenuators set at the same level (the attenuators are wired directly to the board with 3" leads; no pre-amp), the right channel was so far down in volume that the "central" image floated only about 6" or so from the left speaker. I couldn't detect any noteworthy distortion from the right channel, just a reduction in volume that I could almost compensate for by cranking the right volume control to maximum. I then tried switching interconnects at the DSP. I got the same result, and while doing so I noticed a large, sharp turn-on thump from the right speaker (dead silence from the left) followed by a more gradual DC surge (over the course of a couple of seconds or so the cone moved as far forward as the surround allowed and then retreated).
The bias on the funky amp checks out OK, but on rechecking the DC offset I find an off-the-scale initial rise (something above 150mV) followed by a return to the 6mV I measured on both boards initially.
I'll reflow the joints as Chris suggested. If that doesn't correct the problem, what should I look for next? These amps sound so good when they're right I can't stand that they're wrong.
George
AAK said:
Hi, Al:
The thing that first tipped me off that something was wrong was a big imbalance between channels that had not been there previously. With the passive 20K shunt attenuators set at the same level (the attenuators are wired directly to the board with 3" leads; no pre-amp), the right channel was so far down in volume that the "central" image floated only about 6" or so from the left speaker. I couldn't detect any noteworthy distortion from the right channel, just a reduction in volume that I could almost compensate for by cranking the right volume control to maximum. I then tried switching interconnects at the DSP. I got the same result, and while doing so I noticed a large, sharp turn-on thump from the right speaker (dead silence from the left) followed by a more gradual DC surge (over the course of a couple of seconds or so the cone moved as far forward as the surround allowed and then retreated).
The bias on the funky amp checks out OK, but on rechecking the DC offset I find an off-the-scale initial rise (something above 150mV) followed by a return to the 6mV I measured on both boards initially.
I'll reflow the joints as Chris suggested. If that doesn't correct the problem, what should I look for next? These amps sound so good when they're right I can't stand that they're wrong.
George
AndrewT
Performing parts matching to obtain superior performance is a time consuming but eventually fruitful adventure. If you are to do matching are there any statistics as to how many of a given transistor will be needed to obtain the needed sum. If I need 6 devices how many should I expect to purchase to end up with that amount. The 6 figure is arbitrary for example only.
I have no idea where to begin to find this information. Tad
Performing parts matching to obtain superior performance is a time consuming but eventually fruitful adventure. If you are to do matching are there any statistics as to how many of a given transistor will be needed to obtain the needed sum. If I need 6 devices how many should I expect to purchase to end up with that amount. The 6 figure is arbitrary for example only.
I have no idea where to begin to find this information. Tad
Hi Tad,
It depends on the manufacturers process. If there is good control, then your yield will be higher.
For example. Small signal transistors are both very useful and cheap. I normally buy 20 of each type at a time unless they are over priced or expensive. Then I buy 10 of each type. If I want 20 and the price breaks at 25, I will probably buy 25.
For power transistors I may buy 10 at a time. I should get 2 matched pairs out of that. With the On Semi MJW0302A / 0281A I was getting 7 or 8 close matches. Also, NPN's were close to PNP's. A very happy situation. For MJ15024/25 you may get three close or two pairs from ten. MJ15015 / 16 are the other way. NPN's around 60 and PNP's around 180. They are an economy part (says so in the manual = junk).
So you see, it depends on the manufacturer and the part type. You need to buy more fets normally to match than BJT's also.
-Chris
It depends on the manufacturers process. If there is good control, then your yield will be higher.
For example. Small signal transistors are both very useful and cheap. I normally buy 20 of each type at a time unless they are over priced or expensive. Then I buy 10 of each type. If I want 20 and the price breaks at 25, I will probably buy 25.
For power transistors I may buy 10 at a time. I should get 2 matched pairs out of that. With the On Semi MJW0302A / 0281A I was getting 7 or 8 close matches. Also, NPN's were close to PNP's. A very happy situation. For MJ15024/25 you may get three close or two pairs from ten. MJ15015 / 16 are the other way. NPN's around 60 and PNP's around 180. They are an economy part (says so in the manual = junk).
So you see, it depends on the manufacturer and the part type. You need to buy more fets normally to match than BJT's also.
-Chris
How about ageing? Will matched pairs get "divorced" with time?
😉 I seriously consider if e.g. N-channel or P-channel MOS-Fets degenerate in an equal way?
😉 I seriously consider if e.g. N-channel or P-channel MOS-Fets degenerate in an equal way?
That is quite a lot of parts to buy for only a small useable result. Would the mjl4281 transistors fair better in this matching scheme. The pursuit of perfection can be frustrating and expensive. Would you come out better buying say 500 components from a single run as in a group buy. You might have a better chance at obtaining units from the same lot. Tad
Hi all,
I'm ordering my parts and read post 543 about Vbe value.
I'll run with 2X30V transfos... if i use 300r for R24 Vbe will be 3.5(?) .
Is it ok for a good bias with 100r multiturn (i already have).
Thanks.
jf.
I'm ordering my parts and read post 543 about Vbe value.
I'll run with 2X30V transfos... if i use 300r for R24 Vbe will be 3.5(?) .
Is it ok for a good bias with 100r multiturn (i already have).
Thanks.
jf.
Hi Jf,
For 2x30V trans. that's approx. +/-42Vrails. R24 of 300 ohms is to low. I'd recommend for R24 anywhere between 220 ohms and 250 ohms.
Regards,
Al
For 2x30V trans. that's approx. +/-42Vrails. R24 of 300 ohms is to low. I'd recommend for R24 anywhere between 220 ohms and 250 ohms.
Regards,
Al
PS with two voltages
Hi,
I have a PS with +/- 49 VDC for the output and a second voltage (with separate bridge) with +/- 54 VDC, but smaller amperage. Should/could I use it for the input stage, eliminating R31 and R32?
Luis Rossi
Hi,
I have a PS with +/- 49 VDC for the output and a second voltage (with separate bridge) with +/- 54 VDC, but smaller amperage. Should/could I use it for the input stage, eliminating R31 and R32?
Luis Rossi
Hi jef9200,
Hi tryonziess,
Hi The golden mean,
-Chris
Lose the multi turn control.
Hi tryonziess,
I'm sure you will use the TO-92 signal transistors somewhere else with no trouble. They don't cost very much either. For the larger parts, buy 1.5 X what you actually need and hope for the best. On Semi has become very good. The MJL and MJW lines are very good.
Hi The golden mean,
They normally age together although you might want to check them in 20 years. An output stage failure can damage some without causing complete failure. So if you have trouble, recheck these transistors as a precaution. It's very normal to see this with an amplifier that has been "blown".
-Chris
Hi anatech,
Why lose the multiturn? You need it to set bias. Once the bias is set correctly you could measure the pot, and replace it with a voltage divider. But why make things more difficult.
Regards,
Al
Why lose the multiturn? You need it to set bias. Once the bias is set correctly you could measure the pot, and replace it with a voltage divider. But why make things more difficult.
Regards,
Al
Hi Al,
A 270 ° pot is all you need. I noticed your layout after I had bought 6 boards. You don't want to know what I said about that! 😉 Pain in the *** for me now.
Multiturn controls are not reliable for this application. Nor are they required to set the bias where you want it.
-Chris
A 270 ° pot is all you need. I noticed your layout after I had bought 6 boards. You don't want to know what I said about that! 😉 Pain in the *** for me now.
Multiturn controls are not reliable for this application. Nor are they required to set the bias where you want it.
I'm going to bet you will end up with a network of parts there, at least three resistors normally. You will find that the measured value probably is not what you really need also. Just the way life is.
-Chris