yeah... I test it on 2 ohm 680W, and there aint heat at all...link of the measurement again measurement
Passive attenuators with UCD
Hi,
I'm planning to build a passive shunt attentuator as the input source for the UCD 400 modules, and was wondering if there would be any problems driving the UCD from a source with higher than normal output impedence?
The shunt attenuator will have a 2K series resistance in each balanced input and a shunt resistance between the balanced inputs that will vary between 0 and 8K, so the input impedence seen by the UCD module will vary considerably. I will be bypassing the coupling capacitor of the UCD module (in fact plan to have no caps in the signal chain, but will use output DC protection), so variation of frequency response with different volume levels should not be a problem. However it would be good to know that this scheme will not add any problems or distortions.
Regards,
Dean
Hi,
I'm planning to build a passive shunt attentuator as the input source for the UCD 400 modules, and was wondering if there would be any problems driving the UCD from a source with higher than normal output impedence?
The shunt attenuator will have a 2K series resistance in each balanced input and a shunt resistance between the balanced inputs that will vary between 0 and 8K, so the input impedence seen by the UCD module will vary considerably. I will be bypassing the coupling capacitor of the UCD module (in fact plan to have no caps in the signal chain, but will use output DC protection), so variation of frequency response with different volume levels should not be a problem. However it would be good to know that this scheme will not add any problems or distortions.
Regards,
Dean
A question of resistance
Dean,
Both +/- inputs of the module are 100k to ground with just a couple of picofarads thrown in. They are not connected to anything else just the opamp inputs. Assuming you will retain the 8620 opamp and the existing circuit there will be no problem at all. This also assumes you will have the volume control in the same box as the amp and use twisted pair or twisted shielded pair cable to couple in the signal. With the rather high source impedance and much higher input impedance it would be a very good idea to not only shield this cable but to carefully route it away from any high current stuff. I would recommend only grounding the shield at the amp end.
Nice to get to answer one of yours for a change.
Roger
Dean,
Both +/- inputs of the module are 100k to ground with just a couple of picofarads thrown in. They are not connected to anything else just the opamp inputs. Assuming you will retain the 8620 opamp and the existing circuit there will be no problem at all. This also assumes you will have the volume control in the same box as the amp and use twisted pair or twisted shielded pair cable to couple in the signal. With the rather high source impedance and much higher input impedance it would be a very good idea to not only shield this cable but to carefully route it away from any high current stuff. I would recommend only grounding the shield at the amp end.
Nice to get to answer one of yours for a change.
Roger
Thanks Roger.
If it is just straight connect to the opamp noninv and inv inputs then the passive attenuator will work well. Your point about the cable length and shielding is relevant, the attenuator which is a bank of relays with shunt resistors can be mounted near the UCD module. However wont there be a problem without a shield that the EMI from the UCD module or other noise will be introduced into the inputs? With proper layout the CMRR of the opamp will reduce this problem.
What about using low capacitance shielded cable?
The setup I'm planning is:
- Lynx 2B soundcard with CS4396 DAC. No modification to this card is needed as it has an excellent implementation
- Balanced out from the soundcard to the amplifier which has a switch relay, series attenuator resistor (2K Dale for low noise) then 8 x shunt resistors relay switched via PIC between the balanced inputs. To convert this scheme to an unbalanced input simply put another relay to short the -ve balanced lead to earth.
- The attenuator will go straight into the UCD, modified to remove the coupling caps and swap selected caps on board to better types, including the output cap to a Auricap if that checks out with Audience. Change the UCD opamp power supply to one with better regulation.
I'm hoping this will be a "can of woop ***" PC based sound system setup (I hope I got my americanism correct!), ie. extreme UCD....
Regards,
Dean
If it is just straight connect to the opamp noninv and inv inputs then the passive attenuator will work well. Your point about the cable length and shielding is relevant, the attenuator which is a bank of relays with shunt resistors can be mounted near the UCD module. However wont there be a problem without a shield that the EMI from the UCD module or other noise will be introduced into the inputs? With proper layout the CMRR of the opamp will reduce this problem.
What about using low capacitance shielded cable?
The setup I'm planning is:
- Lynx 2B soundcard with CS4396 DAC. No modification to this card is needed as it has an excellent implementation
- Balanced out from the soundcard to the amplifier which has a switch relay, series attenuator resistor (2K Dale for low noise) then 8 x shunt resistors relay switched via PIC between the balanced inputs. To convert this scheme to an unbalanced input simply put another relay to short the -ve balanced lead to earth.
- The attenuator will go straight into the UCD, modified to remove the coupling caps and swap selected caps on board to better types, including the output cap to a Auricap if that checks out with Audience. Change the UCD opamp power supply to one with better regulation.
I'm hoping this will be a "can of woop ***" PC based sound system setup (I hope I got my americanism correct!), ie. extreme UCD....
Regards,
Dean
I am doing what you describe with no great issues. Cheap 10K pot with 10K shunt resistors inline with each balanced leg.
Do note a long way back earlier in this thread that Bruno cautions that this is not a great design since the common mode rejection will be signficantly reduced unless there is very good matching of components in each channel
It's working acceptably here in my situation though...
Do note a long way back earlier in this thread that Bruno cautions that this is not a great design since the common mode rejection will be signficantly reduced unless there is very good matching of components in each channel
It's working acceptably here in my situation though...
With 1% hand picked resistors it should be possible to not significantly reduce the CMRR, which is only appropriate to the series resistors, the shunt between the balanced lines should not matter (as it is common). What I am describing is not separate voltage dividers on each line, but a matched series resistor in each line and a shunt resistor between the two balanced lines.
Regards.
Dean
Regards.
Dean
Resistors,cables and stuff
Chris,
1) Try a higher capacitance cable for the input. This automatically means lower inductance which does have sonic benefits. The higher capacitance will make it harder for the EMI to get in as it sees lower impedance. This also forms a natural R/C filter for RF. The cable will be so short this is not a problem even if the cable capacitance was very high.
2) The best common mode rejection requires absolutely balanced +/- input impedances from DC to several MHz. Balancing the resistors are only part of the problem. You also need to balance all reactive components as well and these will be mostly stray capacitances and inductances so keep leads tightly twisted and the same length.
3) You had asked about the resistors before and I overlooked that there is some benefit to be gained in using .1% parts in the attenuator as well as around the opamp. If this as a lot of expense and work I would say it probably is not worth the effort. The most critical resistors will be the 2k series pass elements and I would recommend the small Caddock MK-132 .1% units available from Michael Percy. (If he only has 1% get 4 and do a best match)
http://www.percyaudio.com
Roger
Chris,
1) Try a higher capacitance cable for the input. This automatically means lower inductance which does have sonic benefits. The higher capacitance will make it harder for the EMI to get in as it sees lower impedance. This also forms a natural R/C filter for RF. The cable will be so short this is not a problem even if the cable capacitance was very high.
2) The best common mode rejection requires absolutely balanced +/- input impedances from DC to several MHz. Balancing the resistors are only part of the problem. You also need to balance all reactive components as well and these will be mostly stray capacitances and inductances so keep leads tightly twisted and the same length.
3) You had asked about the resistors before and I overlooked that there is some benefit to be gained in using .1% parts in the attenuator as well as around the opamp. If this as a lot of expense and work I would say it probably is not worth the effort. The most critical resistors will be the 2k series pass elements and I would recommend the small Caddock MK-132 .1% units available from Michael Percy. (If he only has 1% get 4 and do a best match)
http://www.percyaudio.com
Roger
Roger,
I have a lot of shielded cat5 cable left over from wiring my house with audio (balanced line level signal over shielded twisted pair) and a short length of this should be fine for connecting the balanced attenuator to the UCD. The shield should provide enough capacitance and as you say it gets grounded only at the amp. The quality of cat5 wire, accurate twisting which makes it good for ethernet over twisted pair also makes it good for balanced audio.
Regarding the attenuator for balanced lines, I can't think of a passive scheme that would be better in terms of not upsetting the CMRR than the one I am using (exact matched series resistors with a common shunt between the two balanced lines). If someone has a better scheme please let me know.
Regards,
Dean
I have a lot of shielded cat5 cable left over from wiring my house with audio (balanced line level signal over shielded twisted pair) and a short length of this should be fine for connecting the balanced attenuator to the UCD. The shield should provide enough capacitance and as you say it gets grounded only at the amp. The quality of cat5 wire, accurate twisting which makes it good for ethernet over twisted pair also makes it good for balanced audio.
Regarding the attenuator for balanced lines, I can't think of a passive scheme that would be better in terms of not upsetting the CMRR than the one I am using (exact matched series resistors with a common shunt between the two balanced lines). If someone has a better scheme please let me know.
Regards,
Dean
Hi guys,
I do have a length of CAT5 to sacrifice as well, it will be either that or microphone cable. Supposedly my friend has some high quality mic cable to give me..
All the talk of 1% tolerance is scaring me. Don't forget the CMRR is ruined at higher frequencies by the lesser tolerance of the
"matched" caps used in the feedback loop.
Now if you can find a 1% cap to replace those with, (that is providing they aren't 1% already, I think they might be 5%)... that should help out more in the high range where the ear will be most sensitive?
The idea for a volume control hasn't changed for myself at least, control the volume from the source.
Regards,
Chris
I do have a length of CAT5 to sacrifice as well, it will be either that or microphone cable. Supposedly my friend has some high quality mic cable to give me..
All the talk of 1% tolerance is scaring me. Don't forget the CMRR is ruined at higher frequencies by the lesser tolerance of the
"matched" caps used in the feedback loop.
Now if you can find a 1% cap to replace those with, (that is providing they aren't 1% already, I think they might be 5%)... that should help out more in the high range where the ear will be most sensitive?
The idea for a volume control hasn't changed for myself at least, control the volume from the source.
Regards,
Chris
Cat 5 would have quite high capacitance. Probably low capacitance microphone cable would be better for a passive pre-amp connection.
I'm sure you already know, but the run can be long(er) from the source to the attenuator, but you want to try and minimise the run from the attenuator to the amp
I have only tried my passive attenuator in balanced mode with a single attenuator for each balanced channel. I don't know if the quality would be higher simply switching to single ended or not... It certainly sounds ok like this though.
I'm sure you already know, but the run can be long(er) from the source to the attenuator, but you want to try and minimise the run from the attenuator to the amp
I have only tried my passive attenuator in balanced mode with a single attenuator for each balanced channel. I don't know if the quality would be higher simply switching to single ended or not... It certainly sounds ok like this though.
Cable length
65 CM seems like a lot of unshielded cable to have out there. This is one of those try it and see deals. I don’t think capacitance will be an issue, how noisy your environment is will be though.
Roger
Pretty neat how I can answer before it is asked. Actually I hit the wrong button.
65 CM seems like a lot of unshielded cable to have out there. This is one of those try it and see deals. I don’t think capacitance will be an issue, how noisy your environment is will be though.
Roger
Pretty neat how I can answer before it is asked. Actually I hit the wrong button.
Whats the maximum recommended lead length from Attenuator to module input?
When I get around to ordering a pair of UCD400's I'd like to use my diy boxed 22k stepped Attenuator for the passive pre.
I use multistrand silver pairs which has a PTFE sleeve, the pairs are twisted together for each lead.
The total lead length I'd like to use would measure about 65cm from passive pre to each UCD module, would this be ok?
When I get around to ordering a pair of UCD400's I'd like to use my diy boxed 22k stepped Attenuator for the passive pre.
I use multistrand silver pairs which has a PTFE sleeve, the pairs are twisted together for each lead.
The total lead length I'd like to use would measure about 65cm from passive pre to each UCD module, would this be ok?
Scared?
You are right, I remember from my pro sound days how capacitor tolerances will mess things up. Still shouldn’t we strive to do the best we can in bringing in a clean signal? Switching from 1% to .1 % res. should not be a significant cost unless you insist on using something like the Vishays. The caps are a different story. You used to be able to get 1% polystyrene for pennies. I have no clue about close tolerance SM caps. Is there such a thing and do we even want to go there?
Speaking of close tolerance caps, the new Auricap output filter caps will all be 1%. I am still waiting for the engineering samples to arrive but was told to expect them in less than 2 weeks. I will get a quote on delivery of the first production run next week. Of course the order will not go in till I am sure the product is right.
Roger
Chris,classd4sure said:
You are right, I remember from my pro sound days how capacitor tolerances will mess things up. Still shouldn’t we strive to do the best we can in bringing in a clean signal? Switching from 1% to .1 % res. should not be a significant cost unless you insist on using something like the Vishays. The caps are a different story. You used to be able to get 1% polystyrene for pennies. I have no clue about close tolerance SM caps. Is there such a thing and do we even want to go there?
Speaking of close tolerance caps, the new Auricap output filter caps will all be 1%. I am still waiting for the engineering samples to arrive but was told to expect them in less than 2 weeks. I will get a quote on delivery of the first production run next week. Of course the order will not go in till I am sure the product is right.
Roger
Re: Cable length
Cheers Roger
I can easily use shielded but I've not come across any yet that matches the performance of this stuff I currently use
BTW I'd also recommend those Caddock MK-132 resistors, they are very good and have worked excellent in my dac projects.
The TF020R are even better but not physically suited to all circuits
So many things to try
sx881663 said:
Cheers Roger
I can easily use shielded but I've not come across any yet that matches the performance of this stuff I currently use
BTW I'd also recommend those Caddock MK-132 resistors, they are very good and have worked excellent in my dac projects.
The TF020R are even better but not physically suited to all circuits
So many things to try
I'm not so sure that you won't have problems with high capacitance cables causing an HF rolloff? There are lots of links about this effect on the net. Try perhaps the links from goldpt.com (from memory)
It's easy to figure out what the rolloff will be if you know the capacitance of the cable though
I don't think there is a "maximum" length, but shorter is certainly better (from attenuator to amp). I shouldn't worry so much about the length of interconnect between source and attenuator though.
Maplins do some nice low capacitance mic cable for decent price, and it's shielded with an extra conductor running past the shield (which makes it easy to wire a three way plug)
It's easy to figure out what the rolloff will be if you know the capacitance of the cable though
I don't think there is a "maximum" length, but shorter is certainly better (from attenuator to amp). I shouldn't worry so much about the length of interconnect between source and attenuator though.
Maplins do some nice low capacitance mic cable for decent price, and it's shielded with an extra conductor running past the shield (which makes it easy to wire a three way plug)
Hi,
Ivan, if those are the ones in the feedback path.... yes.
Actually before anyone gets the wrong impression I think I should clarify what my concern actually was.
If I had to guess I'm assuming the tolerance of the given resistors is probably on the order of 0.01% or better??
If it isn't it's because of other tolerance issues making it kind of pointless, I hope.
If they actually are, then replacing them with a part tolerance of .1% wouldn't be an improvement.
I guess we'll need more info, I obviously haven't yet looked into the tolerance of the parts (if it can even be read on the part itself that is).
Cable lengths is pretty standard stuff you can find that on google easily. It will be an issue with single ended setups but if you go the pro audio XLR balanced way it should tolerate alot longer runs.
I also agree that a low capacitance interconnect would be beneficial, also if you start playing with high capacitance cables, length must certainly then become much more of an issue in accordance with its bandwidth.
Cheers,
Chris
Ivan, if those are the ones in the feedback path.... yes.
Actually before anyone gets the wrong impression I think I should clarify what my concern actually was.
If I had to guess I'm assuming the tolerance of the given resistors is probably on the order of 0.01% or better??
If it isn't it's because of other tolerance issues making it kind of pointless, I hope.
If they actually are, then replacing them with a part tolerance of .1% wouldn't be an improvement.
I guess we'll need more info, I obviously haven't yet looked into the tolerance of the parts (if it can even be read on the part itself that is).
Cable lengths is pretty standard stuff you can find that on google easily. It will be an issue with single ended setups but if you go the pro audio XLR balanced way it should tolerate alot longer runs.
I also agree that a low capacitance interconnect would be beneficial, also if you start playing with high capacitance cables, length must certainly then become much more of an issue in accordance with its bandwidth.
Cheers,
Chris