Hi,
I have a Behringer DCX that I have modified with a transformer output stage. It has ~650R output impedance. My amps have an input impedance of 27K.
Now I would ideally like to have an analog volume control but I would rather not add active stages, so I was considering a shunt resistive unit from Welborn Labs. - http://www.welbornelabs.com/remote.htm
My question is, with a 650R output impedance, and 27K input impedance on the amps, is the system going to work okay with a resistive passive volume control or is this likely to mess up the impedance matching? If I used a 50K series resistor in the volume control then I guess the DAC would be happy but the amp would not and vice-versa.
If that is not going to work then I think I would go for a digital volume control, as all my sources are digital. Does anyone know of a unit that can simply take a digital input, dither to 24bit and apply digital attenuation and output a digital signal again for the DAC?
Thanks for your help!
I have a Behringer DCX that I have modified with a transformer output stage. It has ~650R output impedance. My amps have an input impedance of 27K.
Now I would ideally like to have an analog volume control but I would rather not add active stages, so I was considering a shunt resistive unit from Welborn Labs. - http://www.welbornelabs.com/remote.htm
My question is, with a 650R output impedance, and 27K input impedance on the amps, is the system going to work okay with a resistive passive volume control or is this likely to mess up the impedance matching? If I used a 50K series resistor in the volume control then I guess the DAC would be happy but the amp would not and vice-versa.
If that is not going to work then I think I would go for a digital volume control, as all my sources are digital. Does anyone know of a unit that can simply take a digital input, dither to 24bit and apply digital attenuation and output a digital signal again for the DAC?
Thanks for your help!
By impedance matching I meant getting a good match for this purpose, not making them equal.
If I used a 10K pot then the DAC would see 10K which would be just about acceptable but wouldn't that be a little high to drive a 27K amp input. It might work but it seems like its pushing the limits IMO and I might see frequency response variations when the pot is adjusted. A friend has a Welborn labs thing so I might be able to just try it and see.
If I used a 10K pot then the DAC would see 10K which would be just about acceptable but wouldn't that be a little high to drive a 27K amp input. It might work but it seems like its pushing the limits IMO and I might see frequency response variations when the pot is adjusted. A friend has a Welborn labs thing so I might be able to just try it and see.
No, the DAC would see the parallel combination of the lower part of the pot (depends on the setting) and the upper part of the pot in series with the source. At full volume, this would be about the same as the source. At midpoint, it would be about 2.5K. If the input capacitor is too low a value this could affect low frequency response, but I doubt this would be the case.
If it's a concern you could use a 5K pot instead...
If it's a concern you could use a 5K pot instead...
I don't know how you've set up your transformer output - but I've "transformerfied" many a DAC. (My DCX2496 is next).
Voltage output DACs like the AKM in the Behringer like a low load - 1K is often good. As you know, the DAC will see the load reflected back thru the tranfo. The reflected load will depend on the load on the output of the transformer x the winding ratio.
So if you have a 10K load on the secondary side of the transformer with a 1:1 ratio, then the DAC will see 10K. Higher than you want. But if your DAC load resistor is ~ 1.2K, then the DAC will see a total load of 1K. Perfect!
I've used a 10K TKD pot with very, very good results. It means a lower output resistance than a 50K pot.
If you had a six gang 2K pot, you'd be golden.
Hope that makes sense. Basically, use a low value pot and do quick math to be sure you don't load down the DAC too much. No series resistance needed or wanted.
Please ask if this does not make sense.
Voltage output DACs like the AKM in the Behringer like a low load - 1K is often good. As you know, the DAC will see the load reflected back thru the tranfo. The reflected load will depend on the load on the output of the transformer x the winding ratio.
So if you have a 10K load on the secondary side of the transformer with a 1:1 ratio, then the DAC will see 10K. Higher than you want. But if your DAC load resistor is ~ 1.2K, then the DAC will see a total load of 1K. Perfect!
I've used a 10K TKD pot with very, very good results. It means a lower output resistance than a 50K pot.
If you had a six gang 2K pot, you'd be golden.
Hope that makes sense. Basically, use a low value pot and do quick math to be sure you don't load down the DAC too much. No series resistance needed or wanted.
Please ask if this does not make sense.
Hi thanks for the response.
Interesting that you say it likes a low load. On the spec sheet it says to load 600R or greater. I assumed 600 was an absolute minimum and it should be loaded with more.
If the DAC will always see the reflected load, what is the effect of the impedance of the windings themselves? For example how would a 150R : 600R setup differ from 600R : 2K4 ? Is it only of consequence when there is no load connected to the secondary so 600R : 2K4 would better protect the DAC in this case?
I have the transformer in a 1:1 ratio at the moment, and it seems perfectly happy driving a 20K load, and a 10K load.
I just tried a voltage divider of 3k3 (Z1) and 6k6 (Z2) it behaves just fine. I then tried it reversed with 6k6 (Z1) and 3k3 (Z2) and it works fine like that too. This is driving the 10K load of my soundcard input. Great
So how do I calculate the load seen by the DAC, and the source impedance seen by the amp with a voltage divider like this?
In this case would the DAC see Z1 + Z2 as the load? And the amp would see Z2 as the source?
Thanks.
Interesting that you say it likes a low load. On the spec sheet it says to load 600R or greater. I assumed 600 was an absolute minimum and it should be loaded with more.
If the DAC will always see the reflected load, what is the effect of the impedance of the windings themselves? For example how would a 150R : 600R setup differ from 600R : 2K4 ? Is it only of consequence when there is no load connected to the secondary so 600R : 2K4 would better protect the DAC in this case?
I have the transformer in a 1:1 ratio at the moment, and it seems perfectly happy driving a 20K load, and a 10K load.
I just tried a voltage divider of 3k3 (Z1) and 6k6 (Z2) it behaves just fine. I then tried it reversed with 6k6 (Z1) and 3k3 (Z2) and it works fine like that too. This is driving the 10K load of my soundcard input. Great
So how do I calculate the load seen by the DAC, and the source impedance seen by the amp with a voltage divider like this?
An externally hosted image should be here but it was not working when we last tested it.
In this case would the DAC see Z1 + Z2 as the load? And the amp would see Z2 as the source?
Thanks.
When I've talked to the chip designers they usaully say that they spec them at a certain load, e.g. 1K, and don't know how they perform at other loads. There seems to be a wide operating margin, tho.
Don't know why you need a voltage divider, is the output from the DAC too high? I thought the AKM chip in the Behringer output about 1.7V. Is it more?
Or is your voltage divider shown actually the volume pot?
Don't know why you need a voltage divider, is the output from the DAC too high? I thought the AKM chip in the Behringer output about 1.7V. Is it more?
Or is your voltage divider shown actually the volume pot?
Hi,
I didn't have a suitable pot, so to test if it would work well at all, I used a voltage divider from fixed resistors.
So I'm not on the look out for a suitable 6 channel pot. I'd prefer a switched ladder network but I think it will be hard to find one with 6 channels, that would need 12 layers!
Its a shame the Welborn Labs unit is a shunt design as that would be perfect otherwise. You can just stack more boards on it, and its remote control with an LED readout.
I didn't have a suitable pot, so to test if it would work well at all, I used a voltage divider from fixed resistors.
So I'm not on the look out for a suitable 6 channel pot. I'd prefer a switched ladder network but I think it will be hard to find one with 6 channels, that would need 12 layers!
Its a shame the Welborn Labs unit is a shunt design as that would be perfect otherwise. You can just stack more boards on it, and its remote control with an LED readout.
Hi Tenson,
There is more to consider than simply how well an input stage is driven. The impedance of your line (interconnect) and source also has a lot to do with how much noise you pick up. On top of that, if the input is a little non-linear, any resistance in the source will determine how much distortion is created.
I'll be honest about my stance on "passive" preamplifiers. I don't like them. What you are talking about is essentially the same thing here. I feel some people ignore the entire picture and focus on perceived weaknesses. Often, a good buffer is far superior to "no active components in the signal path". Of course, poor buffer designs or poorly executed designs will degrade the signal every time.
I know that an absence of active components in the signal path sounds romantic, but that's all. There is little truth to the statement that the less components the signal goes through, the more "pristine" it is. What is more important is that you account for every problem in a system and deal with it.
Now, the only thing a transformer will buy you is galvanic isolation. This really shouldn't be a problem in a well designed system using intelligently designed devices. The other "feature" that comes along for the ride with a transformer is distortion and reduced bandwidth. This depends on both the quality of the transformer and how well it has been loaded. Transformers need to see the proper impedances on the primary and secondary.
So, which of the two evils would be the better choice?
-Chris
There is more to consider than simply how well an input stage is driven. The impedance of your line (interconnect) and source also has a lot to do with how much noise you pick up. On top of that, if the input is a little non-linear, any resistance in the source will determine how much distortion is created.
I'll be honest about my stance on "passive" preamplifiers. I don't like them. What you are talking about is essentially the same thing here. I feel some people ignore the entire picture and focus on perceived weaknesses. Often, a good buffer is far superior to "no active components in the signal path". Of course, poor buffer designs or poorly executed designs will degrade the signal every time.
I know that an absence of active components in the signal path sounds romantic, but that's all. There is little truth to the statement that the less components the signal goes through, the more "pristine" it is. What is more important is that you account for every problem in a system and deal with it.
Now, the only thing a transformer will buy you is galvanic isolation. This really shouldn't be a problem in a well designed system using intelligently designed devices. The other "feature" that comes along for the ride with a transformer is distortion and reduced bandwidth. This depends on both the quality of the transformer and how well it has been loaded. Transformers need to see the proper impedances on the primary and secondary.
So, which of the two evils would be the better choice?
-Chris
Hi,
I understand your point of view and I don't think one option is naturally superior to the other. With the AKM DAC in the DCX, a transformer is an elegant solution. For one, I don't need to build a complex regulator to optimally power the op-amps. Secondly, the DAC outputs have 2.6V DC on each phase. I believe if I used an active stage I would need to run the summed output through a cap to block the DC? I would still then have to implement a form of volume control.
So in this case I think its nicer to run through a transformer and directly in to a pot. Distortion at 1KHz is around 0.001% so that's low enough for me. Having done a few tests today the frequency response remains linear to about 40KHz on a good range of volume settings. If the TX limits bandwidth beyond that then good, it means there is less high frequency hash getting out of the DAC.
In my post above I meant to type 'I am now looking for a 6 channel pot.
I understand your point of view and I don't think one option is naturally superior to the other. With the AKM DAC in the DCX, a transformer is an elegant solution. For one, I don't need to build a complex regulator to optimally power the op-amps. Secondly, the DAC outputs have 2.6V DC on each phase. I believe if I used an active stage I would need to run the summed output through a cap to block the DC? I would still then have to implement a form of volume control.
So in this case I think its nicer to run through a transformer and directly in to a pot. Distortion at 1KHz is around 0.001% so that's low enough for me. Having done a few tests today the frequency response remains linear to about 40KHz on a good range of volume settings. If the TX limits bandwidth beyond that then good, it means there is less high frequency hash getting out of the DAC.
In my post above I meant to type 'I am now looking for a 6 channel pot.
Hi Tenson,
Yup. I was only suggesting to keep an open mind. There are cases where a good quality transformer is a good solution.
Lastly, the input stage of your power amplifiers are part of the circuit. J FET and tube types are far less of a problem. BJT stages can get a little funny. Experimentation is about the best way to figure that out. Just run your test signal through a 100K pot and measure the THD at different levels (keeping the output of the amp the same). If the THD reading goes up as you increase the series resistance, you need to drive that amp with a low impedance source. Probably less than 1 K ohm would be best in that situation.
You might be just fine. Then again, you may discover that you only thought you were fine. It's your system and no one else can say much unless they duplicate your situation. I certainly can not say anything for certain about your gear and how it interacts.
-Chris
Yup. I was only suggesting to keep an open mind. There are cases where a good quality transformer is a good solution.
I don't worry about midband THD at all. If you have high distortion there, you have problems! It's the frequency extremes that are of interest.
Lastly, the input stage of your power amplifiers are part of the circuit. J FET and tube types are far less of a problem. BJT stages can get a little funny. Experimentation is about the best way to figure that out. Just run your test signal through a 100K pot and measure the THD at different levels (keeping the output of the amp the same). If the THD reading goes up as you increase the series resistance, you need to drive that amp with a low impedance source. Probably less than 1 K ohm would be best in that situation.
You might be just fine. Then again, you may discover that you only thought you were fine. It's your system and no one else can say much unless they duplicate your situation. I certainly can not say anything for certain about your gear and how it interacts.
-Chris
Hi,
I'm not sure how you mean I should measure the disto. Could you explain in a little more detail how to setup?
My power amps are SKA GB150, if that helps.
Edited to add:
I'm not arguing in favor of any design here, but it might interest you to know I just did a small disto test at 100Hz to compare the transformer output of the DCX to the op-amp output of my soundcard. Its an M-Audio Fast Track Ultra, so its a good card and I trust them to design a good bit of kit.
Green is the transformer output, Red is the soundcards op-amp output. It is interesting to note that while the transformer has clearly higher lower order harmonics, it has clear 7th and 9th harmonics, while the transformer output has nothing above the noise floor beyond the 5th harmonic. The soundcard however has practically no 4th harmonic, while the transformer does. So, which is 'better' at low frequencies is not really as clear as one might hope from a THD figure.
Also interesting is that the ADC (soundcard input) is clearly low enough distortion to reveal all of this!
I'm not sure how you mean I should measure the disto. Could you explain in a little more detail how to setup?
My power amps are SKA GB150, if that helps.
Edited to add:
I'm not arguing in favor of any design here, but it might interest you to know I just did a small disto test at 100Hz to compare the transformer output of the DCX to the op-amp output of my soundcard. Its an M-Audio Fast Track Ultra, so its a good card and I trust them to design a good bit of kit.
Green is the transformer output, Red is the soundcards op-amp output. It is interesting to note that while the transformer has clearly higher lower order harmonics, it has clear 7th and 9th harmonics, while the transformer output has nothing above the noise floor beyond the 5th harmonic. The soundcard however has practically no 4th harmonic, while the transformer does. So, which is 'better' at low frequencies is not really as clear as one might hope from a THD figure.
Also interesting is that the ADC (soundcard input) is clearly low enough distortion to reveal all of this!
Damn I just missed the editing cut off time before I could upload the picture.
Having done some more testing the TX output seems even more favorable at 1KHz where it is about equal or lower at low harmonics and far superior at higher harmonics than the soundcards opamp output, or the DEQ (non-modified) output I also have here.
I'm starting to understand why I like the sound of these transformer modified outputs
An externally hosted image should be here but it was not working when we last tested it.
Having done some more testing the TX output seems even more favorable at 1KHz where it is about equal or lower at low harmonics and far superior at higher harmonics than the soundcards opamp output, or the DEQ (non-modified) output I also have here.
I'm starting to understand why I like the sound of these transformer modified outputs
Hi Tenson,
I haven't been that happy using a sound card. It does give you some information, but I'm not sure it's that accurate. Who knows?
I generally use either a Spectrum Analyzer or a THD meter, often I use both. What you do is input a low distortion sine wave and set your level to where you want it. Then measure the output. Note that you are often using a fixed frequency for doing this "the old school way". I can't afford the newer test gear. The THD meter (HP 339A) removes the fundamental frequency and measures what is left. The published residual on this tester is 0.0018%. You can also use a Spectrum Analyzer to sweep your frequency range. I use an older HP 3585A for that. You can also connect the Spectrum Analyzer to the output of the THD meter to extend your dynamic range.
Your 100 Hz test shows a lot, try lower frequencies too. Measure the output before the transformer too so you can eliminate your "THD floor". With a sound card, trying to measure distortion at 10 KHz or 20 KHz will not be possible really. You might only see the second or third harmonic. This is where the older HP THD meters (331A, 332A, 333A, 334A or 339A). The 8903A is more scarce and expensive. You would need an oscillator with all of these except the 339A and 8903A as they contain their own.
I hope some of that is some help to you.
-Chris
I haven't been that happy using a sound card. It does give you some information, but I'm not sure it's that accurate. Who knows?
I generally use either a Spectrum Analyzer or a THD meter, often I use both. What you do is input a low distortion sine wave and set your level to where you want it. Then measure the output. Note that you are often using a fixed frequency for doing this "the old school way". I can't afford the newer test gear. The THD meter (HP 339A) removes the fundamental frequency and measures what is left. The published residual on this tester is 0.0018%. You can also use a Spectrum Analyzer to sweep your frequency range. I use an older HP 3585A for that. You can also connect the Spectrum Analyzer to the output of the THD meter to extend your dynamic range.
Your 100 Hz test shows a lot, try lower frequencies too. Measure the output before the transformer too so you can eliminate your "THD floor". With a sound card, trying to measure distortion at 10 KHz or 20 KHz will not be possible really. You might only see the second or third harmonic. This is where the older HP THD meters (331A, 332A, 333A, 334A or 339A). The 8903A is more scarce and expensive. You would need an oscillator with all of these except the 339A and 8903A as they contain their own.
I hope some of that is some help to you.
-Chris
Thanks for your reply. I can not really invest in all that equipment just for this so my soundcard will have to do.
The bit I was not sure of was how I can have the output connected to my amp, and yet also measure the disto at the same time? If I parallel connected it to the souncard and amp then the load would be effected perhaps making the test useless?
I might be able to connect the souncard via some very high value resistors so it doesn't effect the output and amp input combo?
The bit I was not sure of was how I can have the output connected to my amp, and yet also measure the disto at the same time? If I parallel connected it to the souncard and amp then the load would be effected perhaps making the test useless?
I might be able to connect the souncard via some very high value resistors so it doesn't effect the output and amp input combo?
Hi Tenson,
The accepted practice would be to connect your amplifier output to an 8 or 4 ohm "dummy load". A high powered resistor. A 5, 6, 7.5 or 10 ohm resistor would do really. Make sure it's mounted where it won't burn anything and is rated for 20 watts or better. The "real" dummy loads are like the ones Dale makes. Mine are 8 ohm, 250 watt non-inductive and mounted on heat sinks. I had to buy mine for my repair shop, I do not expect most people to do that. If I were to replace mine today, I would buy the hollow 225 watt type close to 8 ohms and pad it to reach the proper value. 7.5 ohm is attractive as you could correct for wire and connections.
You must insert series and parallel resistances to your sound card to keep from blowing it sky high. I think mine measured close to 10 K, so your pad would be 10:1 or 100:1 to keep the input voltages from exceeding your sound card specs. Figure some lower resistance across the sound card input, like 1.1K maybe. This in parallel with your sound card should be roughly 1 K, then you adjust your series resistor to provide the proper voltage drop. This resistance will not affect your amplifier in any way.
This would be a tool you will get a lot of use from I think.
-Chris
The accepted practice would be to connect your amplifier output to an 8 or 4 ohm "dummy load". A high powered resistor. A 5, 6, 7.5 or 10 ohm resistor would do really. Make sure it's mounted where it won't burn anything and is rated for 20 watts or better. The "real" dummy loads are like the ones Dale makes. Mine are 8 ohm, 250 watt non-inductive and mounted on heat sinks. I had to buy mine for my repair shop, I do not expect most people to do that. If I were to replace mine today, I would buy the hollow 225 watt type close to 8 ohms and pad it to reach the proper value. 7.5 ohm is attractive as you could correct for wire and connections.
You must insert series and parallel resistances to your sound card to keep from blowing it sky high. I think mine measured close to 10 K, so your pad would be 10:1 or 100:1 to keep the input voltages from exceeding your sound card specs. Figure some lower resistance across the sound card input, like 1.1K maybe. This in parallel with your sound card should be roughly 1 K, then you adjust your series resistor to provide the proper voltage drop. This resistance will not affect your amplifier in any way.
This would be a tool you will get a lot of use from I think.
-Chris
Hi,
Sorry I thought you meant that I should measure the output of the DAC, while it is connected to the amp, as the amps input may effect this if it is not a resistive load.
I have just done an IMD test to compare the transformer outputs, the soundcard and non-modified DEQ output. Now it gets a bit more interesting. The soundcard output, while it had some noticeable higher harmonic disto, has practically no IMD at all! Very impressive. The Transformer output has quite obvious IMD. The stock DEQ output on the other hand has even worse IMD than either of these. Green is the transformer output, Red is the soundcard and Purple is the stock DEQ.
It just goes to show that an op-amp or transformer do not always have higher disto or lower IMD. Implementation and personal compromises are everything.
Given that I want to build a 6 channel volume control in to this, and IMD is often more audible than harmonic disto I am starting to agree that a buffered design may be best for me. Now I need your advice on how to build a simple and capable buffered volume control that will have vanishing levels of IMD and low high order HD.
Something based on the new National op-amp maybe? When it comes to feedback I have no idea what I'm doing.
Sorry I thought you meant that I should measure the output of the DAC, while it is connected to the amp, as the amps input may effect this if it is not a resistive load.
I have just done an IMD test to compare the transformer outputs, the soundcard and non-modified DEQ output. Now it gets a bit more interesting. The soundcard output, while it had some noticeable higher harmonic disto, has practically no IMD at all! Very impressive. The Transformer output has quite obvious IMD. The stock DEQ output on the other hand has even worse IMD than either of these. Green is the transformer output, Red is the soundcard and Purple is the stock DEQ.
It just goes to show that an op-amp or transformer do not always have higher disto or lower IMD. Implementation and personal compromises are everything.
An externally hosted image should be here but it was not working when we last tested it.
Given that I want to build a 6 channel volume control in to this, and IMD is often more audible than harmonic disto I am starting to agree that a buffered design may be best for me. Now I need your advice on how to build a simple and capable buffered volume control that will have vanishing levels of IMD and low high order HD.
Something based on the new National op-amp maybe? When it comes to feedback I have no idea what I'm doing.
Back to OT (Original Topic), exact impedance match is desired for 2 reasons:
1. Power transfer for less losses that means less amplification factor is needed so less noises/distortions are generated. With modern SS gear it is not the reason since input and output resistances are frequency depended and non- linear, so mismatch is desired (higher loar resistance for lower output resistance, in order to get less distortions on non-linear impedances)
2. Impedance match with long lines. Also, it is not an issue since the length of the cable is much less than 1/4 of wavelength of the highest frequency in the cable.
Transformer: it's main significance is balance for best interference cancellation. No matter how well your gear designed, you will always have ground loops when connect more than one peaces of gear since you must ground each chassis through the power outlet.
Also, where transformers are needed, when you have to deliver low output impedance from a gear powered by weak current supply, the typical case is phantom power (3.4 kiloohm resistance, 48V) of the microphone that must drive 150 Ohm inputs through long cables.
1. Power transfer for less losses that means less amplification factor is needed so less noises/distortions are generated. With modern SS gear it is not the reason since input and output resistances are frequency depended and non- linear, so mismatch is desired (higher loar resistance for lower output resistance, in order to get less distortions on non-linear impedances)
2. Impedance match with long lines. Also, it is not an issue since the length of the cable is much less than 1/4 of wavelength of the highest frequency in the cable.
Transformer: it's main significance is balance for best interference cancellation. No matter how well your gear designed, you will always have ground loops when connect more than one peaces of gear since you must ground each chassis through the power outlet.
Also, where transformers are needed, when you have to deliver low output impedance from a gear powered by weak current supply, the typical case is phantom power (3.4 kiloohm resistance, 48V) of the microphone that must drive 150 Ohm inputs through long cables.
Hi Tenson,
Look into using a buffer after your volume control. Something along the lines of a "diamond buffer" would work well. A simple circuit, but you can buy the BUF634. These are fast devices, so bypassing is important.
Hi Anatoliy,
-Chris
Look into using a buffer after your volume control. Something along the lines of a "diamond buffer" would work well. A simple circuit, but you can buy the BUF634. These are fast devices, so bypassing is important.
Read the data sheet and any app notes. This buffer does not use overall negative feedback.
Absolutely! That is one of the points I made earlier.
Hi Anatoliy,
Correct, but this is not this situation. The signal is strong and we are trying to retain the signal power.
True again, but also not an issue in this case. Home music system.
That's on topic? Your example is specifically a live or studio recording gig. It will not be seen in normal home systems. I think you are defending transformer use in general while we are examining this particular situation.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.