No, it would be 40mv plus the offset generated by the missing inputcap.
This is because of the DC-blocking cap in the feedback, which drops gain to 1 for dc.
Mike
This is because of the DC-blocking cap in the feedback, which drops gain to 1 for dc.
Mike
OK, finished 6 channels now. Feels like industrial production...
The DC offsets are very good and it seems this is real.
On three channels I get unmeasurable DC offsets: 0.0 or -0.2 mV, in that range. On two more channels it is about 2 to 3 mV. On the last channel I made a mistake - picked the wrong input transistors, where the Ic was matched but the hfe was off by 4% (the others were really good matches, sometimes no significant difference with my DMM). I thought the desoldering in the tight layou might do more harm than good and left them in. Plus here on the output transistors I was left with less perfect matches, maybe 5-6% mismatch, still quite good actually. Well, this last channel has 10 mV offset. So I think the zero offset channels are for real.
So, matching does really matter quite a bit and I sure got lucky with a batch that must have come from the exact same wafer.
In HD I can't get "real" numbers, I can just see that the channels perform the same at 16v RMS into 6R, leading with 2nd at -74 dB (0.02%) and decreasing 3rd and 5th harmonics, 4th is absent.
But even these numbers are likely "generated" by my soundcard setup. I finally measured the soundcard for in and output wiring and it turns out even the unbalanced output has 100R to ground, and both out and inputs are AC coupled. The outputs have an additional 100k to ground as bleeder I suppose, and the balanced out has a floating sleeve. So that makes it completely impossible for me to get input, output and DUT to run at the same commons, and since I can't input a balanced signal to the Symasym I am out of luck with that equipment. There will always be a frequency dependent common mode offset. It is hum and noise resistant and resilient but not quite made for measurement.
BTW I wonder what our other PCB buyers are doing? We seem to be a lonely bunch here, Ryssen, Cromodora and myself 😉 .
The DC offsets are very good and it seems this is real.
On three channels I get unmeasurable DC offsets: 0.0 or -0.2 mV, in that range. On two more channels it is about 2 to 3 mV. On the last channel I made a mistake - picked the wrong input transistors, where the Ic was matched but the hfe was off by 4% (the others were really good matches, sometimes no significant difference with my DMM). I thought the desoldering in the tight layou might do more harm than good and left them in. Plus here on the output transistors I was left with less perfect matches, maybe 5-6% mismatch, still quite good actually. Well, this last channel has 10 mV offset. So I think the zero offset channels are for real.
So, matching does really matter quite a bit and I sure got lucky with a batch that must have come from the exact same wafer.
In HD I can't get "real" numbers, I can just see that the channels perform the same at 16v RMS into 6R, leading with 2nd at -74 dB (0.02%) and decreasing 3rd and 5th harmonics, 4th is absent.
But even these numbers are likely "generated" by my soundcard setup. I finally measured the soundcard for in and output wiring and it turns out even the unbalanced output has 100R to ground, and both out and inputs are AC coupled. The outputs have an additional 100k to ground as bleeder I suppose, and the balanced out has a floating sleeve. So that makes it completely impossible for me to get input, output and DUT to run at the same commons, and since I can't input a balanced signal to the Symasym I am out of luck with that equipment. There will always be a frequency dependent common mode offset. It is hum and noise resistant and resilient but not quite made for measurement.
BTW I wonder what our other PCB buyers are doing? We seem to be a lonely bunch here, Ryssen, Cromodora and myself 😉 .
Hi Andrew,
When FETs or complementary differential inputs are used, I still would use an input coupling capacitor. If the input impedance is high, you can use a poly type capacitor. Much less destructive to sound quality.
-Chris
Yes. I often calculate the expected DC offset to make sure everything is running properly. Most designs have an inherent DC offset.
When FETs or complementary differential inputs are used, I still would use an input coupling capacitor. If the input impedance is high, you can use a poly type capacitor. Much less destructive to sound quality.
-Chris
Channel 2 is running..
Offset is -5,5 to -0,1mv, never steady moving all the time.Why?
The other channel is now:between 8-12mv moving around to..
Bias is now 85mv at R27/28.
Offset is -5,5 to -0,1mv, never steady moving all the time.Why?
The other channel is now:between 8-12mv moving around to..
Bias is now 85mv at R27/28.
Ryssen,
I noticed that bias and DC offset will fluctuate very easily due to temperature variations on the drivers. Since they're freestanding, there is no large heatsink "buffer" and things such as moving my hand close to the amp to attach a test lead will make them move substantially. When I first tried to set bias I had a fan on in the room - forget it: plus minus 30% when the fan blew over the amp.
Conditions should be more stable inside a casing.
I noticed that bias and DC offset will fluctuate very easily due to temperature variations on the drivers. Since they're freestanding, there is no large heatsink "buffer" and things such as moving my hand close to the amp to attach a test lead will make them move substantially. When I first tried to set bias I had a fan on in the room - forget it: plus minus 30% when the fan blew over the amp.
Conditions should be more stable inside a casing.
first real listen
OK, so I finished mounting my completed six channels into my speakers (each speaker has onboard analog EQ and XO, and 3 channels of amplification), replacing the LM3886 I had previously there. Due to space constraints I am running 3 channels with a single 225 VA XF for the time being. With triamping and near 10000 uF / channel it should be alright anyway.
The mounting was quite an ordeal because I had to reorganize the grounding scheme from previously earthed to earth lifted (antiparallel diodes) including the preamp. Not quite done with it yet because I still get pops at startup and bangs when muting the preamp now, I presume that's the different grounds adjusting when they are connected/disconnected.
So, how does it sound?
Actually, surprise: little spectacular difference at first to the LM3886's, maybe I had those quite optimized after all. Couldn't test at high SPLs yet, at normal levels what stood out was increased detail and depth. No trace of "phaseyness" for lack of a better word, no trace of grain in the highs, and depending on material, either silky smooth or with bite. Timbre, punch, dynamics etc, couldn't spot much of a difference. So I guess that means they're what they're supposed to be: neutral. Will report more when I have opportunity for more critical listening.
Some questions:
- when I tested the amps on the bench I blew an amp fuse on two channels on two occasions, due to massize ground loop into a load. But the other rail fuse didn't blow, resulting in 2.5V DC on the output. Is this expected, and if so, what is the point of the fuses - 2.5V DC indefinitely will sure melt any speaker at some point.
- reportedly Symasym has no turn on or off noises. Mine do have a small turn on bump, followed by strong but rapidly decreasing hum (like an overloaded transformer, but coming from the speaker). All in all maybe 1 second of "thud-hum-silence". Could this be caused by the 0.2R at the DC input of the Capacitor banks of each channel? My transformer is very close, but in its own aluminium box.
w/o conections the amps are extremely qiet btw, even on my 96 dB/W woofers. With XO connected there is some hum and hiss but this comes from the XO: patched in air wires and grounding issues of its own. I just had not noticed it so much with the 26 dB gain LM3886's.
OK, so I finished mounting my completed six channels into my speakers (each speaker has onboard analog EQ and XO, and 3 channels of amplification), replacing the LM3886 I had previously there. Due to space constraints I am running 3 channels with a single 225 VA XF for the time being. With triamping and near 10000 uF / channel it should be alright anyway.
The mounting was quite an ordeal because I had to reorganize the grounding scheme from previously earthed to earth lifted (antiparallel diodes) including the preamp. Not quite done with it yet because I still get pops at startup and bangs when muting the preamp now, I presume that's the different grounds adjusting when they are connected/disconnected.
So, how does it sound?
Actually, surprise: little spectacular difference at first to the LM3886's, maybe I had those quite optimized after all. Couldn't test at high SPLs yet, at normal levels what stood out was increased detail and depth. No trace of "phaseyness" for lack of a better word, no trace of grain in the highs, and depending on material, either silky smooth or with bite. Timbre, punch, dynamics etc, couldn't spot much of a difference. So I guess that means they're what they're supposed to be: neutral. Will report more when I have opportunity for more critical listening.
Some questions:
- when I tested the amps on the bench I blew an amp fuse on two channels on two occasions, due to massize ground loop into a load. But the other rail fuse didn't blow, resulting in 2.5V DC on the output. Is this expected, and if so, what is the point of the fuses - 2.5V DC indefinitely will sure melt any speaker at some point.
- reportedly Symasym has no turn on or off noises. Mine do have a small turn on bump, followed by strong but rapidly decreasing hum (like an overloaded transformer, but coming from the speaker). All in all maybe 1 second of "thud-hum-silence". Could this be caused by the 0.2R at the DC input of the Capacitor banks of each channel? My transformer is very close, but in its own aluminium box.
w/o conections the amps are extremely qiet btw, even on my 96 dB/W woofers. With XO connected there is some hum and hiss but this comes from the XO: patched in air wires and grounding issues of its own. I just had not noticed it so much with the 26 dB gain LM3886's.
Hi MBK,
Try mounting reverse biased rectifiers from the circuit side of each fuse to ground. This will prevent supply reversal when a fuse blows (to a maximum of 0.8 VDC typically).
-Chris
Edit:
Try mounting reverse biased rectifiers from the circuit side of each fuse to ground. This will prevent supply reversal when a fuse blows (to a maximum of 0.8 VDC typically).
-Chris
Edit:
This sounds like a DC offset that is correcting itself. It could also be HF oscillation that dies down as the circuit stabilizes. You need to watch the speaker outputs with a DC meter and an oscilloscope to figure out what is happening.
DC offset definitely goes from say 50 mV to the usual 0 - 3 mV within a short time, seconds maybe, as bias climbs.
Edit: this from watching both DC offset and bias behavior at startup in real time on 2 DMMs at DC setting (but the sampling rates of the DMMs are rather low). DC offset starts high and settles in 1-3 seconds. Turn on thump hard to measure correctly with the DMMs, it is too short. The woofers move quite a bit though, several mm, even the 88 dB/W mid.
Scope, yes, the missing link...
Edit: this from watching both DC offset and bias behavior at startup in real time on 2 DMMs at DC setting (but the sampling rates of the DMMs are rather low). DC offset starts high and settles in 1-3 seconds. Turn on thump hard to measure correctly with the DMMs, it is too short. The woofers move quite a bit though, several mm, even the 88 dB/W mid.
Scope, yes, the missing link...
Hi Andrew,
Hi MBK,
Can you change the sampling rate on your DVM. Some meters will allow you to drop resolution to trade for speed. Your DC offset would have to be much higher to do what you have reported.
Why not make an RF detector probe for your meter? All it will tell you really is whether you do in fact have oscillation or not. Also, check out the temperature of your zobel resistor. It may get warm (finger tip test).
-Chris
They certainly are.
Hi MBK,
Can you change the sampling rate on your DVM. Some meters will allow you to drop resolution to trade for speed. Your DC offset would have to be much higher to do what you have reported.
Why not make an RF detector probe for your meter? All it will tell you really is whether you do in fact have oscillation or not. Also, check out the temperature of your zobel resistor. It may get warm (finger tip test).
-Chris
Hi Chris,
Will look into this. It's a Craftsman though, not a Fluke 😉 (I am aware of that other thread!) . One time I think I saw a 2V and once a 13V number flashing by (DC offset), for fractions of a second, that would sound more like the woofer movement, next reading around 50 mV, slowly decaying to 0 - 3 mV, all this in 1 - 3 seconds.
Now that would be great, how do I go about this? I dimly remember seeing an article using a Germanium diode I believe. I sometimes use the child monitor (basically a walkie talkie receiver) as a primitive detector but it is too sensitive, the hash from the nearby transformer bridge is enough to get you plenty noise there (!!).
Zobel, when I had the amps on a bench I often checked, never warm (even with the floating heatsink problems). Now mounted in the box direct access to the PCB is a lot harder to impossible unfortunately.
anatech said:
Will look into this. It's a Craftsman though, not a Fluke 😉 (I am aware of that other thread!) . One time I think I saw a 2V and once a 13V number flashing by (DC offset), for fractions of a second, that would sound more like the woofer movement, next reading around 50 mV, slowly decaying to 0 - 3 mV, all this in 1 - 3 seconds.
anatech said:
Now that would be great, how do I go about this? I dimly remember seeing an article using a Germanium diode I believe. I sometimes use the child monitor (basically a walkie talkie receiver) as a primitive detector but it is too sensitive, the hash from the nearby transformer bridge is enough to get you plenty noise there (!!).
Zobel, when I had the amps on a bench I often checked, never warm (even with the floating heatsink problems). Now mounted in the box direct access to the PCB is a lot harder to impossible unfortunately.
Hi MBK,
For measuring your start up DC offset, can you lock your meter to the 20V scale? This will save time in autoranging. Still, integrating a changing signal will give you a very inaccurate answer.
I am totally spoiled with my HP 34401A. They still sell for $$$$$$$ on Ebay used!
An inexpensive pointer type meter may be your best friend right now. No good for high impedance circuits, but fine on an amplifier output. You can try it with your made up RF probe. Try and "Google" the phrase.
-Chris
That's the idea. A silicon diode followed by a small capacitor and a resistor to ground. Measure across the resistor on your DC scale.
For measuring your start up DC offset, can you lock your meter to the 20V scale? This will save time in autoranging. Still, integrating a changing signal will give you a very inaccurate answer.
I am totally spoiled with my HP 34401A. They still sell for $$$$$$$ on Ebay used!
An inexpensive pointer type meter may be your best friend right now. No good for high impedance circuits, but fine on an amplifier output. You can try it with your made up RF probe. Try and "Google" the phrase.
-Chris
Hi Chris,
This HP meter looks superb indeed. Well, they also have plenty of old oscilloscopes 2nd hand in store here, around 200 USD for the 20 MHz models, w/o probe. I am always on the verge of breaking down for one, I just don't really trust what they mean with "working condition" and I don't want another riddle to solve: is it the scope that's not working or is it me that doesn't know how to use it? haha. not.
For sure I saw old style analog multimeters in the stores, new, for cheap, so that part is easy. RF probe: so a small signal Silicon diode will also work? Thought it needed to be Germanium.
BTW I saw in another thread you have trouble finding M3 taping screws and 2.5 mm bits over there. Well I bought my taps, box of 3, Japanese, for around 2 USD here, and the 2.5mm bit (Black and Decker, they do know metric!) for another, err, 1 USD or so. PM me if interested, this stuff doesn't weigh much and should stand in for the beer I offered earlier. 😀
To Ryssen, AAK, Cromodora and all others: do you also get a turn on thump? or is it just me?
This HP meter looks superb indeed. Well, they also have plenty of old oscilloscopes 2nd hand in store here, around 200 USD for the 20 MHz models, w/o probe. I am always on the verge of breaking down for one, I just don't really trust what they mean with "working condition" and I don't want another riddle to solve: is it the scope that's not working or is it me that doesn't know how to use it? haha. not.
For sure I saw old style analog multimeters in the stores, new, for cheap, so that part is easy. RF probe: so a small signal Silicon diode will also work? Thought it needed to be Germanium.
BTW I saw in another thread you have trouble finding M3 taping screws and 2.5 mm bits over there. Well I bought my taps, box of 3, Japanese, for around 2 USD here, and the 2.5mm bit (Black and Decker, they do know metric!) for another, err, 1 USD or so. PM me if interested, this stuff doesn't weigh much and should stand in for the beer I offered earlier. 😀
To Ryssen, AAK, Cromodora and all others: do you also get a turn on thump? or is it just me?
No real "thumb",when connect the mains,I get a hum,not to loud,wich slowly faids away 1-3 seconds,until it is "quiet".
OK that reassures me, sounds like what I am getting, except on a 96 dB/W woofer it is kinda audible. 🙄
Hi MBK,
I get a barely audible thump on power up. I've never tried using the MPSA18 for an input, but, it's possible using the 2N5551 instead could be causing it. The MPSA18 has a much higher Hfe.
Al
I get a barely audible thump on power up. I've never tried using the MPSA18 for an input, but, it's possible using the 2N5551 instead could be causing it. The MPSA18 has a much higher Hfe.
Al
Thanks Al, well my batch of 2N5551 measured a meager 100-130 Hfe, same as the drivers and output devices actually.
Hi MBK,
The low hFE could cause this. Your gains do seen like they are very low. You are in a very good area to pick up Japanese semi's (real ones). 2SA970 and 2SC2240 are favorite devices of mine. Very, very popular in amplifiers. If you redo the pinout, they would work just fine in there. They are also good to 120 V C-E. 😀
Can you get large quantities for <$?> ? Look at 100 lots in a BL gain ranking. The data sheets are PDF's over our posting limit. PM me and I'll send them to you.
Magura may send me a 3M tap and drill (2.5 mm). However, I sometimes do a lot of work. I break things if I'm too tired and too stubborn to go to sleep. If you can check some of these out for me I should pay you. Do you have a PayPal account?
There may be some other Japanese part numbers that would suit this project well. I'm thinking driver transistors. Other favorites of mine are 2SC3421 and 2SA1358, 2SC3423 and 2SA1360, 2SC3467 and 2SA1370. There are more. These are old classics and I have to find new TO-220 drivers. On Semi has been making some new outputs that are finally well worth using. Let's hope they flesh out their line to cover drivers and smaller signal devices.
I see some really good deals on 'scopes on Ebay in your area. For $200 you should be able to pick up a 100 MHz scope. Understand that they will all need work by now. Don't be tempted to buy some of the inexpensive digital 'scopes. With a DSO you need a ton more performance to make it as useful as an analog 20 MHz 'scope. That means $$$$$$$$.
-Chris
The low hFE could cause this. Your gains do seen like they are very low. You are in a very good area to pick up Japanese semi's (real ones). 2SA970 and 2SC2240 are favorite devices of mine. Very, very popular in amplifiers. If you redo the pinout, they would work just fine in there. They are also good to 120 V C-E. 😀
Can you get large quantities for <$?> ? Look at 100 lots in a BL gain ranking. The data sheets are PDF's over our posting limit. PM me and I'll send them to you.
Magura may send me a 3M tap and drill (2.5 mm). However, I sometimes do a lot of work. I break things if I'm too tired and too stubborn to go to sleep. If you can check some of these out for me I should pay you. Do you have a PayPal account?
There may be some other Japanese part numbers that would suit this project well. I'm thinking driver transistors. Other favorites of mine are 2SC3421 and 2SA1358, 2SC3423 and 2SA1360, 2SC3467 and 2SA1370. There are more. These are old classics and I have to find new TO-220 drivers. On Semi has been making some new outputs that are finally well worth using. Let's hope they flesh out their line to cover drivers and smaller signal devices.
I see some really good deals on 'scopes on Ebay in your area. For $200 you should be able to pick up a 100 MHz scope. Understand that they will all need work by now. Don't be tempted to buy some of the inexpensive digital 'scopes. With a DSO you need a ton more performance to make it as useful as an analog 20 MHz 'scope. That means $$$$$$$$.
-Chris
MBK said:
Sorry. Have not sorted amp out yet.
Current status:
Yet to install new pair of BD139s. Brand new black ones from Farnell to replace grey surplus ones bought from you know where.
Also just finished rebuilding Carlos' AD815pre.
This and the Symasym will form a one box integrated. No nulling circuit. If I read that thread correctly, can do without as the Symasym has caps at input. Only setback is the input DC will eat up volume pot wipers.
Mulling over:
While the heatsink is unmounted, I'm thinking of changing the feedback cap to Oscon 470uF/10v. Also, while at it why not I change R5 and R6 from generic to better types. These degen resistors greatly determine noise level of input pair right?