Hi,
The pops are not really that bad, but we do want to eliminate them. Muting is silent (no popping), so that may be an option.
Petter,
I am having a strange problem with impedences (I think).
I am using the APOX-1 using the 5K resistors that Craig specified. The APOX-1 is at the output of a XBOSOZ which goes into an Aleph2. At cetain volume settings I get a noticaeable 60Hz hum. At most settings, none. For instance, at 169, none, 170 has moderate hum. 171-186 no hum, but at 187 high hum. Would this be related to a high Zin or high Zout?
Grataku,
I modified the wiring when installing the APOX-1 to make sure the channels were not running in parallel. What a difference. Not just a "I guess I hear a difference". The imaging tightened and the soundstage widened considerably. I will run the crosstalk test today. I suspect that it is the wiring and not the differences between the APXO-1/Apox2?
For instance the dogs in "Amused to Death" moved from just right of the speakers to the next county.
Dale
The pops are not really that bad, but we do want to eliminate them. Muting is silent (no popping), so that may be an option.
Petter,
I am having a strange problem with impedences (I think).
I am using the APOX-1 using the 5K resistors that Craig specified. The APOX-1 is at the output of a XBOSOZ which goes into an Aleph2. At cetain volume settings I get a noticaeable 60Hz hum. At most settings, none. For instance, at 169, none, 170 has moderate hum. 171-186 no hum, but at 187 high hum. Would this be related to a high Zin or high Zout?
Grataku,
I modified the wiring when installing the APOX-1 to make sure the channels were not running in parallel. What a difference. Not just a "I guess I hear a difference". The imaging tightened and the soundstage widened considerably. I will run the crosstalk test today. I suspect that it is the wiring and not the differences between the APXO-1/Apox2?
For instance the dogs in "Amused to Death" moved from just right of the speakers to the next county.
Dale
Attachments
I buildt this shunt volume control from aleph P 1.0 with ADC0804. Of corse there are a lot of pops when turning. Since then i think about substitution with a MC. I thought of why this occures. I think its because there is some time when tuning volume the relais contacts fly from one position to the other and no relais is closed. Would it be a solution to time correct the relais so a close signal is earlyer sent than an open signal, and switch not all relais that should close first, but close one, wait, open one, close one etc ... to make the process of switching softer? So always from est to should state with the lowest jumps in resistance.
I have some input here which might be useful in the idea generation phase for Apox-1 switching scheme:
1. Check out specifications for relay close/opening times. I have looked in my catalogue for various relays, and they appear to be all over the place in whether they break before make or vice versa. As you say, pass relays should be break before make, and the others typically the other way round.
2. When switching pass relays, one is doing some pretty radical level shifting. I thing one should consider how often one has to switch, and definitely not switch more than one range at a time (unless you use your pretty cool method for doing it).
3. A way reduce the swithoff time for a relay is to use a zener diode up to 27 or so volts instead of the ordinary switchoff diode. I know you guys are currently using diodes in the drivers, so this might be a mute point unless you can find drivers that do not have diodes built-in. It is possible that the switching noise can itself be a function of crosstalk from driver to relay -- there is one long relay driver path that runs in parallell to one of the channel signals for the length of the board. When this relay switches off, it will have a very long path back to the driver which will spike since there is no diode at the relay itself. You might want to try shorting this out with a wire. The wires in question are: IC3 pin 17 and the positive relay voltage as well. K9 is potentially one relay affected.
6. Another way to make a relay switch off quicker is to add a secondary current path so that to set it, current passes through a resistor which is low enough to keep it at set once latched, and a capacitor+resistor combination which will add the latching force but obviously return to zero current as cap is charged up. The final way I can think of to modify a relay's switching speed either on or off is to use a diode in paralell with a resistor. The direction of the diode will determine whether "on" or "off" is adjusted.
7. Consider replacing pass element relays (I am assuming they are the potentially the biggest problem), or add localized free-wheeling diodes to them so that they are at least matched for shutoff.
8. The mute relay could potentially be replaced by a solid state unit (which will typically switch an order or two faster than an ordinary unit) which would operate really quickly and one-shot to zero between switching levels. The basic design problem here is that it needs to be bidirectiona (bidirectional solid state relays are available) unless one is willing to switch to ground. It is also possible to switch in a capacitor to low-pass filter when changing volume. If doing that, it should not be a big problem shorting to ground, and one could do what NP does -- saturate an ordinary cheap BJT, and it will conduct both ways, but you don't know at what level it is (and with capacitor in series (as low-pass filter), it won't really matter to what voltage level one switches. If using a solid state relay, the mute or a low significant bit can be cycled every time volume changes, and timing will be "automatic".
9. The base resistance position can be used to host a semiconductor based switched low-pass filter as noted above (8). It could be controlled from one of the free outputs, and it could be optically controlled if necessary.
Regardless of what is going on, try a progression which is on the same pass element at all times and see if you get the same pops or not.
Now to the important bit: What sounds best -- is it Apox-1 or Apox-2?
Petter
1. Check out specifications for relay close/opening times. I have looked in my catalogue for various relays, and they appear to be all over the place in whether they break before make or vice versa. As you say, pass relays should be break before make, and the others typically the other way round.
2. When switching pass relays, one is doing some pretty radical level shifting. I thing one should consider how often one has to switch, and definitely not switch more than one range at a time (unless you use your pretty cool method for doing it).
3. A way reduce the swithoff time for a relay is to use a zener diode up to 27 or so volts instead of the ordinary switchoff diode. I know you guys are currently using diodes in the drivers, so this might be a mute point unless you can find drivers that do not have diodes built-in. It is possible that the switching noise can itself be a function of crosstalk from driver to relay -- there is one long relay driver path that runs in parallell to one of the channel signals for the length of the board. When this relay switches off, it will have a very long path back to the driver which will spike since there is no diode at the relay itself. You might want to try shorting this out with a wire. The wires in question are: IC3 pin 17 and the positive relay voltage as well. K9 is potentially one relay affected.
6. Another way to make a relay switch off quicker is to add a secondary current path so that to set it, current passes through a resistor which is low enough to keep it at set once latched, and a capacitor+resistor combination which will add the latching force but obviously return to zero current as cap is charged up. The final way I can think of to modify a relay's switching speed either on or off is to use a diode in paralell with a resistor. The direction of the diode will determine whether "on" or "off" is adjusted.
7. Consider replacing pass element relays (I am assuming they are the potentially the biggest problem), or add localized free-wheeling diodes to them so that they are at least matched for shutoff.
8. The mute relay could potentially be replaced by a solid state unit (which will typically switch an order or two faster than an ordinary unit) which would operate really quickly and one-shot to zero between switching levels. The basic design problem here is that it needs to be bidirectiona (bidirectional solid state relays are available) unless one is willing to switch to ground. It is also possible to switch in a capacitor to low-pass filter when changing volume. If doing that, it should not be a big problem shorting to ground, and one could do what NP does -- saturate an ordinary cheap BJT, and it will conduct both ways, but you don't know at what level it is (and with capacitor in series (as low-pass filter), it won't really matter to what voltage level one switches. If using a solid state relay, the mute or a low significant bit can be cycled every time volume changes, and timing will be "automatic".
9. The base resistance position can be used to host a semiconductor based switched low-pass filter as noted above (8). It could be controlled from one of the free outputs, and it could be optically controlled if necessary.
Regardless of what is going on, try a progression which is on the same pass element at all times and see if you get the same pops or not.
Now to the important bit: What sounds best -- is it Apox-1 or Apox-2?
Petter
Hum questions
Dale,
Is it 60Hz or 120Hz hum. Most people refer to 120Hz hum as 60Hz hum, and they are the result of different sources (bridge rectified and non-rectified power). If it is 120Hz, consider attempting to power Apox with batteries to verify performance of power supply.
I am looking at the C program output, and note that there should not really be a big difference between position 186 and 187 since they use the same pass resistor. It should be instructive to check what happens between them, as this might assist finding the biggest problem. I don't have time to check out the bit pattern, but hope it is not K9 .... (re point 3 in previous post) which could generate 120Hz noise, in which case one resolution could be to improve the digital power supply or cut the trace at both ends and feed with a piece of wire. Please check out the sequence {3,27}, {3,28}, {3,29} to verify what is going on 🙂
However, it appears that you are certainly not in top-dog territory with either of these.
Craig, may I suggest you publish the entire sequence (all 1020) with bit patterns per relay? This should enable more eyes to be able to look and assist in coming up with ideas.
If the worst comes to the worst, it might be necessary to listen to all relevant position and exclude them from the sequence. Software is great, isn't it?
Petter
Dale,
Is it 60Hz or 120Hz hum. Most people refer to 120Hz hum as 60Hz hum, and they are the result of different sources (bridge rectified and non-rectified power). If it is 120Hz, consider attempting to power Apox with batteries to verify performance of power supply.
I am looking at the C program output, and note that there should not really be a big difference between position 186 and 187 since they use the same pass resistor. It should be instructive to check what happens between them, as this might assist finding the biggest problem. I don't have time to check out the bit pattern, but hope it is not K9 .... (re point 3 in previous post) which could generate 120Hz noise, in which case one resolution could be to improve the digital power supply or cut the trace at both ends and feed with a piece of wire. Please check out the sequence {3,27}, {3,28}, {3,29} to verify what is going on 🙂
However, it appears that you are certainly not in top-dog territory with either of these.
Craig, may I suggest you publish the entire sequence (all 1020) with bit patterns per relay? This should enable more eyes to be able to look and assist in coming up with ideas.
If the worst comes to the worst, it might be necessary to listen to all relevant position and exclude them from the sequence. Software is great, isn't it?
Petter
Hi Petter,
What does that mean. The performance of the APOX?
I will go look at the various signal with a scope and see what frequency the noise is at.
So far, the sound of the APOX1 is extremely good (at volumes with no hum).
Whether it is my change in wiring or the differences in the APOX-1 vs APOX-2, the sound right now is really amazing.
Dale
Top dog:
I fully expect both Apox-1 and APox-2 to be the top dogs in terms of performance. I have already voted with my dollars and went for Apox-1 over Apox-2 for various reasons (including anticipated performance between the two and usability etc.). While I have not gotten it yet, I fully expect the performance to be a cut above most if not every offering on the market.
What I was referring to in these terms top dog was referenced to a prior posting I did in terms of sequencing where my theory was that "as much "power" as possible should be transferred to next stage (where the phrase "only top two top dogs can play" was coined if it was necessary to obtain a higher number of levels than if one purely goes for "only top dog can play") which means that a low series impedance should be used at high levels.
So I am sorry about the confusion I have created. This was surely unintended.
Petter
I fully expect both Apox-1 and APox-2 to be the top dogs in terms of performance. I have already voted with my dollars and went for Apox-1 over Apox-2 for various reasons (including anticipated performance between the two and usability etc.). While I have not gotten it yet, I fully expect the performance to be a cut above most if not every offering on the market.
What I was referring to in these terms top dog was referenced to a prior posting I did in terms of sequencing where my theory was that "as much "power" as possible should be transferred to next stage (where the phrase "only top two top dogs can play" was coined if it was necessary to obtain a higher number of levels than if one purely goes for "only top dog can play") which means that a low series impedance should be used at high levels.
So I am sorry about the confusion I have created. This was surely unintended.
Petter
Dale and Craig,
How about separate boards for each channel of the Apox1? I suppose you can do it now and leave out one channel but at the expense of extra components.
It would be great if you could perform seperation measurements.
Regards,
Jam
How about separate boards for each channel of the Apox1? I suppose you can do it now and leave out one channel but at the expense of extra components.
It would be great if you could perform seperation measurements.
Regards,
Jam
Hi Petter,
No offense was taken at all. I just wasn't clear about the reference.
We will not ship out until we are happy with the results!!!
I am now looking at the supplies.
I am using a switching 5V supply. I will change to a linear regulated and see what happens. I do see the noise on the scope. It does turn on/off at the values that I posted.
JAM,
I will do some crosstalk tests today. Craig and I did some yesterday, and we found that we are at least -60dB down. Unfortunately, that is the limit of our meter.
I will do the listening test.
Dale
No offense was taken at all. I just wasn't clear about the reference.
We will not ship out until we are happy with the results!!!
I am now looking at the supplies.
I am using a switching 5V supply. I will change to a linear regulated and see what happens. I do see the noise on the scope. It does turn on/off at the values that I posted.
JAM,
I will do some crosstalk tests today. Craig and I did some yesterday, and we found that we are at least -60dB down. Unfortunately, that is the limit of our meter.
I will do the listening test.
Dale
jam said:
It would be a great idea to make APOX 1 on separate boards for dual mono configurations as well for having separate modules for Home Theater setups (uneven number of channels)
Change to battery if you want to make sure that the power supply is totally out of the path (for reference) to verify it is not from there that any hum might be coming 🙂
I too am interested in the notion of Apox-1 mono-boards. I have seen some rather nice SIL relays, and I am sure the board could be made pretty damn small. Perhaps so small that it can be used to plug directly into existing pot-configurations (vertically perhaps).
If you do decided to go with mono boards, please call me (not that I think todays performance is suboptimal, just that I think it would be even easier to integrate monos in a system, particularly if they were low profile with input and output on same side so they could go in as a daughtercard 🙂)
If you want, I would be willing to do layout work and prototyping for you.
And by the way, a mono version should be on teflon ....
Petter 🙂
I too am interested in the notion of Apox-1 mono-boards. I have seen some rather nice SIL relays, and I am sure the board could be made pretty damn small. Perhaps so small that it can be used to plug directly into existing pot-configurations (vertically perhaps).
If you do decided to go with mono boards, please call me (not that I think todays performance is suboptimal, just that I think it would be even easier to integrate monos in a system, particularly if they were low profile with input and output on same side so they could go in as a daughtercard 🙂)
If you want, I would be willing to do layout work and prototyping for you.
And by the way, a mono version should be on teflon ....
Petter 🙂
Get rid of it 😉
Whilst this may not be the source of the problem, I'd definitely get it out of the equation.
SMPS supplies, unless you're really clever, will always be noisier than the most basic linear supply, and the EMI problems need to be dealt with too.
Certainly for evaluation I'd stick to linear.
I'm getting tempted to get my cheque book out already 🙂
Andy.
Whilst this may not be the source of the problem, I'd definitely get it out of the equation.
SMPS supplies, unless you're really clever, will always be noisier than the most basic linear supply, and the EMI problems need to be dealt with too.
Certainly for evaluation I'd stick to linear.
I'm getting tempted to get my cheque book out already 🙂
Andy.
harvardian said:
Dale,
I am not worried about anything being 'wrong' with the setup, I came to my own conclusions about the apox system, the reason I haven't paid for it yet is that I will pay for it in cash. I have a couple of $thou in the house for the emergencies and this qualifies as one, if I put another $500 on the visa my wife is going to deny me the already scarce 'affections' she has toward me. That's really unfair actually, I shouldn't have said that. 😉
I don't know if you already did this...remove everything and in essence use the apox as a passive preamp. Switch the gain stage to be after the APOX. Put the transformers on the GND.
This is a general question for which I don't know the answer: is coupling the XLR + and - via the shunt resistors a good idea? Petter had probably a lot to do with this in trying to save on a couple of rows of resistors...😉
Hi Guys,
A mono APOX-1 would be no problem. Yes, the relays would only need to be 1 form A and there are numerous options.
Our only concern in changing components is the sourcing/stock issues. Also, the more variations that we have may increase the cost, since it would dilute the quantities.
I put in a very clean 5V linear supply (no batteries) and get the same result.
I think that it is in my other components. At the non-hum settings, there is no hum at all. The signals look extremely clean.
Dale
A mono APOX-1 would be no problem. Yes, the relays would only need to be 1 form A and there are numerous options.
Our only concern in changing components is the sourcing/stock issues. Also, the more variations that we have may increase the cost, since it would dilute the quantities.
I put in a very clean 5V linear supply (no batteries) and get the same result.
I think that it is in my other components. At the non-hum settings, there is no hum at all. The signals look extremely clean.
Dale
Dale,
I really doubt its the +5V supply.
Its much more likely to be coming in from the big Victoria Magnetics transformers. Your +/-60 Volt regulators are probably the most likely cause. They were running very hot at idle.
Maybe they are on the edge. Can you put a scope on the +/-60V
supply, and AC couple your scope, so you can see if the DC supplies are passing the noise.
These regulators are probably the most likely source of the noise
coupling through your X-BOSOZ.
-Craig
I really doubt its the +5V supply.
Its much more likely to be coming in from the big Victoria Magnetics transformers. Your +/-60 Volt regulators are probably the most likely cause. They were running very hot at idle.
Maybe they are on the edge. Can you put a scope on the +/-60V
supply, and AC couple your scope, so you can see if the DC supplies are passing the noise.
These regulators are probably the most likely source of the noise
coupling through your X-BOSOZ.
-Craig
Shiled plane
Have you tried disconnecting, or changing the connections to, the shield plane.
It's something to eliminate...
Additionally I'd like to highlight Petter's comment about the back-EMF diodes.
These should be directly across the relays, and may be part of the popping problem.
The problem is that the very long track running power to the relays (that runs right through the physical audio path) will be radiating vast, relative to signal levels, EMI as the relays switch off. I accept this is only a problem when switching, but is less than ideal.
A diode and / or a snubbing network may work wonders here.
The diodes in the drivers are there primarily to protect the o/p switch and are too far away to kill the EMI.
Andy.
Have you tried disconnecting, or changing the connections to, the shield plane.
It's something to eliminate...
Additionally I'd like to highlight Petter's comment about the back-EMF diodes.
These should be directly across the relays, and may be part of the popping problem.
The problem is that the very long track running power to the relays (that runs right through the physical audio path) will be radiating vast, relative to signal levels, EMI as the relays switch off. I accept this is only a problem when switching, but is less than ideal.
A diode and / or a snubbing network may work wonders here.
The diodes in the drivers are there primarily to protect the o/p switch and are too far away to kill the EMI.
Andy.
Another thought
It may be worth trying the shield plane connected to the 5V digital supply ground.
Whilst this may add digital noise in the plane, it will immediately reduce the loop areas dramatically for the relay supplies / driver feeds.
It may prove something...
Andy.
It may be worth trying the shield plane connected to the 5V digital supply ground.
Whilst this may add digital noise in the plane, it will immediately reduce the loop areas dramatically for the relay supplies / driver feeds.
It may prove something...
Andy.
grataku said:
Touche! 🙂
My position is that a balanced input following Apox will per definition remove common mode noise which is what I believe is referred to in this post by my friend Grataku. Differential mode noise will likely be better attenuated by using a direct between phase method. It is true that if there is common mode on the signal, the following stage needs to work harder to remove it, but getting a CMR of 30+dB is pretty easy these days so I don't consider this relevant unless there is substantial CM on the lines (in which case there is probably more to gain from doing work elsewhere than move to relay switched volume control).
Now, if the input following the Apox is single ended, the result will be the same since one shorts the one phase to ground anyway.
Now, Grataku, I certainly know you are having some fun, and I like that. Still, the reason for wanting to shunt between phases was twofold, and even though it has been discussed deeply on this forum it is perhaps timely to revisit the three main anticipated benefits which are likely to enhance sonics (in my book):
a) Allow operation without having to worry about DC level of signal which allows for extremely simple 2 stage (one before, the other after the Apox) gain stages with full DC coupling.
b) The wish to have as few relays in series with the signal as possible.
c) half the inductance of the shunts, courtesy of aforementioned resistor saving "technology" 🙂
I don't understand why there is a need to attenuate to ground at all. Indeed, my friends locally are always messing with me about my signal grounding philosophy (which is don't use ground unless you absolutely need it) so I do appreciate that people have their own ways. Then again saving resistors is not a bad idea either 🙂
Petter
By my earlier suggestion of "putting the transformers on the GND" I meant remove them from the vicinity of the Apox circuit and put them on the floor.
Disconnecting the gainstage and just using the apox1 as a passive volume control should pretty much tell where the noise is coming from.
Can the the EMI produced by a 5VDC relay be _that_ severe?
The diodes across the relay coils are probably good practice anyway.
Disconnecting the gainstage and just using the apox1 as a passive volume control should pretty much tell where the noise is coming from.
Can the the EMI produced by a 5VDC relay be _that_ severe?
The diodes across the relay coils are probably good practice anyway.
Have you tried turning the transformers in your pre-amp? Typically, the place where the wires exit have the most "hum" coming out of them.
And since nobody said how beatiful the '1 was, perhaps I will step up and do that very thing 🙂
Petter
(who has access to Teflon double sided boards and prototyping equipment ...)
And since nobody said how beatiful the '1 was, perhaps I will step up and do that very thing 🙂
Petter
(who has access to Teflon double sided boards and prototyping equipment ...)
- Status
- Not open for further replies.