I very much would appreciate some one to prove – either right or wrong – what I found strange when measuring such line transformers some years ago.
I digged up the following plots showing different distortion figures of the same line transformer with just different source impedance due to no / big / small capacitor:
driven without capacitor (-58dB distortion)
driven with 100uF capacitor (-57dB distortion)
driven with 10uF capacitor (-49dB distortion)
###########
Measurements were taken at output of small (matchbox size) 1:1 no-name transformer @
0.76 Vrms
source impedance ~ 50 Ohm from SSM2142
Load ~ 25 kOhm from SSM2143
0 dB on plots equals ~760 mVrms (wrong scaling on plots !)
############
Strange thing I found here is, that distortion seems to substantially increase with rising source impedance ???
Above would indicate that we should better be pretty carefully with the source impedance (the lower the better) ?
Any comments ?
Michael
I digged up the following plots showing different distortion figures of the same line transformer with just different source impedance due to no / big / small capacitor:
driven without capacitor (-58dB distortion)
driven with 100uF capacitor (-57dB distortion)
driven with 10uF capacitor (-49dB distortion)
###########
Measurements were taken at output of small (matchbox size) 1:1 no-name transformer @
0.76 Vrms
source impedance ~ 50 Ohm from SSM2142
Load ~ 25 kOhm from SSM2143
0 dB on plots equals ~760 mVrms (wrong scaling on plots !)
############
Strange thing I found here is, that distortion seems to substantially increase with rising source impedance ???
Above would indicate that we should better be pretty carefully with the source impedance (the lower the better) ?
Any comments ?
Michael
Last edited:
Well I was not really able to determine the output impedance of the AKM 4393 in the usual way. I used a 5K pot as a load slowly reducing the resistance. Hoped to find the 1/2 voltage point.
But did not. Voltage stays constant until the load drops to about 80 ohms, then it clips badly. Lower than that results in the waveform going nuts. Maybe short circuit protection in the chip. I tested at 40Hz, 1Khz and 6.3Khz all -9dBFS with exactly the same results.
So I'm not sure what the actual output impedance is, but maybe we can call it 100 ohms?
But did not. Voltage stays constant until the load drops to about 80 ohms, then it clips badly. Lower than that results in the waveform going nuts. Maybe short circuit protection in the chip. I tested at 40Hz, 1Khz and 6.3Khz all -9dBFS with exactly the same results.
So I'm not sure what the actual output impedance is, but maybe we can call it 100 ohms?
Hi Pano
What you are seeing when looking at the outputs of a voltage output DAC are the outputs of a pair of OPAMPs with feedback. I suspect that the output impedance is very low, less than .1 ohm. The feedback around the I-V conversion in the DAC keeps the output voltage constant until you hit the current limit of the output stage. Looking at the AKM4393 DAC they list the max output current at 3.5ma. They also list 1K ohm as the min load for a DC load and 600 ohms for a min AC load. I'm assuming that they mean 1K to ground per phase or 600 ohms between the phases.
Were you connecting the load between the output phases or to ground?
The way that most of the voltage out DACs have a DC coupled load makes one wonder about if the output stage is running class A? Also makes one wonder if the DC load resistors should remain even if you are driving the primary of a transformer to keep the bias current up in the output stage. On your many different transformer output stages have you left the DC load in place or removed them?
More questions than answers...
Gary
What you are seeing when looking at the outputs of a voltage output DAC are the outputs of a pair of OPAMPs with feedback. I suspect that the output impedance is very low, less than .1 ohm. The feedback around the I-V conversion in the DAC keeps the output voltage constant until you hit the current limit of the output stage. Looking at the AKM4393 DAC they list the max output current at 3.5ma. They also list 1K ohm as the min load for a DC load and 600 ohms for a min AC load. I'm assuming that they mean 1K to ground per phase or 600 ohms between the phases.
Were you connecting the load between the output phases or to ground?
The way that most of the voltage out DACs have a DC coupled load makes one wonder about if the output stage is running class A? Also makes one wonder if the DC load resistors should remain even if you are driving the primary of a transformer to keep the bias current up in the output stage. On your many different transformer output stages have you left the DC load in place or removed them?
More questions than answers...
Gary
I very much would appreciate some one to prove – either right or wrong – what I found strange when measuring such line transformers some years ago.
snip
###########
Measurements were taken at output of small (matchbox size) 1:1 no-name transformer @
0.76 Vrms
source impedance ~ 50 Ohm from SSM2142
Load ~ 25 kOhm from SSM2143
0 dB on plots equals ~760 mVrms (wrong scaling on plots !)
############
Strange thing I found here is, that distortion seems to substantially increase with rising source impedance ???
Above would indicate that we should better be pretty carefully with the source impedance (the lower the better) ?
Any comments ?
Michael
Hi Michael,
What you are seeing is normal. Almost all transformers will have increasing distortion with increasing drive impedance. If you carefully look at specifications provided by Lundahl, Sowter, and others you will note that the source impedance will be as low as they can get it to show low distortion. Also note that for showing the best frequency response the source impedance may be different than the impedance used for lowest distortion. In cases where a transformer has a resonance at the high end of the frequency range the HF peak will be worse with low source impedance.
Gary
I agree that this format is more difficult to read than the spectrum analyzer output but it contains much more data in a single chart. It would take many charts from the spectrum analyzer to come close to the detail with frequency as a single swept chart. I'm not sure what caused the bumpy even harmonics but I'll see if anything jumps out...
Taking the ideas presented so far and doing some more thinking (danger! danger!) I'm leaning towards 2 voltages- 2Vrms and .2Vrms for testing the transformers. 20Vrms would be nice for transformers that are for interstage use in tube amps, but I don't have a buffer that can do that (yet). For impedance I would like to test at 100 ohms, 300 ohms, and 1K (maybe 3K for tube interstage transformers). For a load on the secondary I think that 20K would be a good compromise
I'll see if I have any tests that will measure the impedance automatically. Too many runs to do by hand.
At this point we are at 5 or 6 (7 or 8 if impedance is done) charts for each transformer. The test only takes 30 seconds to run so it won't be too bad, just lots of images to keep track of. All the data about the test configuration can be included in the text string at the bottom of the image.
To do the different input impedance checks I'll have to modify the output resistors in the M-Audio card in my lab computer. The stock breakout resistors are 150 ohms each I have some 27.4 ohm 1206 SMT resistors I'll drop into the circuit and check out. If all is good I'll setup a 2 pole switch to add the other resistors in line for the higher impedance settings.
Onto another measurement project...
I picked up one of my brother's spare stock Behringer DCX2496 crossovers last weekend. I'm planning on comparing his stock unit to my unit with the alternate input circuit I've been using for a couple of years. I was using 1:1 output transformers directly driven from the DAC for awhile until I acquired a Crown comtech 800 amp for sub duties. After that I was running the DAC directly to both the main and sub amps as they have balanced inputs and did not care about the DC offset.
It will be interesting to see how they compare for noise and distortion when measured in the same test setup.
Gary
Hey Gary, I believe you are right about the opamp outputs of the chip. I'll try to measure current into 1K tomorrow. It may or may not tell me anything.
My tests were all done in differential mode - so the resistor was between the +/- pins and not to ground. The way a transformer would be connected. I do always leave a resistive load across the DAC output, more to flatten the impedance than any thing else. But that's not a DC load. Might have to try that.
I have done impedance measurements on a number of transformers. I just use the WT3 woofer tester. It maxes out at 10K, so I have to run the tests at a 9K ohms secondary load or less. That still shows a great deal of difference between transformers. I can supply these graphs if needed.
My tests were all done in differential mode - so the resistor was between the +/- pins and not to ground. The way a transformer would be connected. I do always leave a resistive load across the DAC output, more to flatten the impedance than any thing else. But that's not a DC load. Might have to try that.
I have done impedance measurements on a number of transformers. I just use the WT3 woofer tester. It maxes out at 10K, so I have to run the tests at a 9K ohms secondary load or less. That still shows a great deal of difference between transformers. I can supply these graphs if needed.
Pano, Gary – good thoughts and results so far!
Please do not forget to point out about some cable impedance as load too. Cable runs in my setup are quite often around 10m or more to get the amps close to speakers – I assume this to be the case for many other too...
Could be done that way – but suffers from precision in calculation when output impedance gets veeeery low – as seems the case from Pano's measurements
I strongly agree that the sweep format is a looooot more informative than the single frequency measurements I've shown (they stem from "pre-ARTA" age
)
Doing so many plots to check for a variety of applications would really be *great* !
Thanks a lot for confirmation .
I was aware of that conflict regarding distortion versus FR optimisation but have not found that to be outspoken – so thought my measurements might have been wrong.
A question worth to be explored I agree – a simple THD run with open out versus resistors to ground measurement might be enough to determine though
Michael
Please do not forget to point out about some cable impedance as load too. Cable runs in my setup are quite often around 10m or more to get the amps close to speakers – I assume this to be the case for many other too...
Could be done that way – but suffers from precision in calculation when output impedance gets veeeery low – as seems the case from Pano's measurements
I strongly agree that the sweep format is a looooot more informative than the single frequency measurements I've shown (they stem from "pre-ARTA" age
)Doing so many plots to check for a variety of applications would really be *great* !
Thanks a lot for confirmation .
I was aware of that conflict regarding distortion versus FR optimisation but have not found that to be outspoken – so thought my measurements might have been wrong.
A question worth to be explored I agree – a simple THD run with open out versus resistors to ground measurement might be enough to determine though
Michael
Last edited:
Thinking about – I changed my mind.
This is an issue more about tweaking FR in the upper department and to a lesser extent about distortion – so, probably should better be kept apart as for a separate chapter of line transformers.
Michael
Yes, I'd rather see the load as either a pure 20K, or something approaching a typical 2 meter cable and amp input section.
Don't know about this whole DC load thing. I've always tried to avoid it. I suppose I could run a load from the primary center tap to ground. That would be both an AC and DC load - equal on both halves. Not sure what effect that would have on the output stage of the DAC chip.
Don't know about this whole DC load thing. I've always tried to avoid it. I suppose I could run a load from the primary center tap to ground. That would be both an AC and DC load - equal on both halves. Not sure what effect that would have on the output stage of the DAC chip.
OK, did some AC current measurements into 5K and 1K load.
Both seem to show that the output impedance of the AKM is ~100 ohms.
Maybe a 50R resistor on each leg inside the chip? That's my best guess.
Also: Max level is 1.77V RMS straight out of the chip into 1K. Thats is just as the clip LED comes on.
Both seem to show that the output impedance of the AKM is ~100 ohms.
Maybe a 50R resistor on each leg inside the chip? That's my best guess.
Also: Max level is 1.77V RMS straight out of the chip into 1K. Thats is just as the clip LED comes on.
Quick question: can LM4562 op-amps be installed straight to DCX in the place of old op-amps? Have somebody tried them? LM4562's specs look pretty damn good, definitely more capable op-amps than originals. I like the low input noise (only 2.7nV/ √ ^Hz (typ)) as it reduces the hiss(?).
Could someone write a little summary of desoldering/resoldering of SOIC op-amps? How it is done the most convenient way (is there such)?
Could someone write a little summary of desoldering/resoldering of SOIC op-amps? How it is done the most convenient way (is there such
)?
Last edited:
yes they can.
to desolder, just use some wick or a desoldering pump along with your iron. then solder the 4562s in using a pair of tweezers and some flux.
here is a SMD soldering tutorial. YouTube - Surface Mount Soldering 101
have fun with your improved fidelity.
to desolder, just use some wick or a desoldering pump along with your iron. then solder the 4562s in using a pair of tweezers and some flux.
here is a SMD soldering tutorial. YouTube - Surface Mount Soldering 101
have fun with your improved fidelity.
Not sure about the compatibility of the op amps, but the most common approach for general SMD rework is to use hot air for desoldering.
A special hot air gun is used to heat the leads of the part, and the part can then be removed with tweezers once everything's hot enough. The pads can then be cleaned up with solder wick.
When resoldering, I use liquid flux. With a good dose of liquid flux, even the finest leads can be (relatively) easily soldered with a fine-tip iron and a tiny bit of solder. The liquid flux allows the solder to flow really well, so if there's too much solder, you can clean the iron tip and then draw off the extra solder with it. After everything's soldered, the flux can be cleaned off with alchohol and an acid brush... blow dry with compressed air.
Hejet makes a nice hot air gun, if you can find one (narrow tip and cooling air all around). Weller also makes a good, relatively inexpensive, one.
A special hot air gun is used to heat the leads of the part, and the part can then be removed with tweezers once everything's hot enough. The pads can then be cleaned up with solder wick.
When resoldering, I use liquid flux. With a good dose of liquid flux, even the finest leads can be (relatively) easily soldered with a fine-tip iron and a tiny bit of solder. The liquid flux allows the solder to flow really well, so if there's too much solder, you can clean the iron tip and then draw off the extra solder with it. After everything's soldered, the flux can be cleaned off with alchohol and an acid brush... blow dry with compressed air.
Hejet makes a nice hot air gun, if you can find one (narrow tip and cooling air all around). Weller also makes a good, relatively inexpensive, one.
Have you, personally, done this?
I worked for many years as an electronics bench tech. I've replaced 100 pin QFP's and all sorts of other SMD's, I've even resolderd leads in the middle of flexible PCBs, but I don't think I'd stand a chance lifting even an 8-pin SOIC with a desoldering pump.
Laying down a bunch of wick and using it to heat the whole side of an SOIC, and then prying up that one side... maybe. But a pump? If you have the chip cold with any solder under the legs, the legs will still be soldered to the board, no pump or wick is going to pull all the solder out from under a foot. You need to have all the leads on at least one side hot enough that the solder is molten so that you can pull up that one side.
Worst worst case, you could use a very thin iron to bend up each leg, one at a time, if you didn't care about the part you were removing (like in this case).