The PCB has 4 holes for socket pins whereas sockets that tie pin 1 direct to the chassis only have 3. So it just means one through hole is not used, that's all. It bothers me
Oh, and the only socket I can see from Neutrik which does connect pin 1 direct to chassis in it's male form is the B series. A series version doesn't seem to exist any more. It is available from Amphenol however in an A series version, fully compatible with Neutrik.
I guess having the separate ground through hole does mean you can use either type, so perhaps that's why he made it that way.
As I mentioned before, Enrico's 6 input board only has 3 holes for its sockets and so means that only one type can be used so anyone looking to move over to his 6 input board at a later date may be better off doing the same as me else they won't be able to re-use their 4 pin sockets. - actually, forget that because his uses the vertical form anyway..
Right, am going to go ahead with my order using a different type of socket to the original BOM and leaving the separate ground holes on the PCB unused.. all sockets will share ground via pin 1 and the groundplane.
Last edited:
The PCB has 4 holes for socket pins whereas sockets that tie pin 1 direct to the chassis only have 3. So it just means one through hole is not used, that's all. It bothers me
Sorry if I caused any confusion.
The PCB has mounting holes for almost every option. In those case where the connector's mounting hole is not directly connected to pin 1, there is a connection between Pin1 and the chassis ground via the ground plane of the PCB.
So for the male contact use the MC3MAH. For this connector said connection between pin1 and ground plane on the PCB will connect to the chassis via the separate ground pin and the contact in the mounting hole.
For the female connector use the Neutrik NC3FAH1 . In this socket pin1 is directly connected to Pin1, but it also connects through the groundplane and the separate ground connectors of the connector to the chassis via the mounting hole.
Hans
I made it simple for myself.
I used the XLR sockets I had in stock.
All were the panel mounting type.
Socket is electrically connected to the back panel via the filed metal hole in the panel and via the 2 mounting screws+nuts.
Pin1 is connected to the shell tag.
Pins 2 & 3 connect via a twisted pair to the signal pads on the PCB. Simples.
I used the XLR sockets I had in stock.
All were the panel mounting type.
Socket is electrically connected to the back panel via the filed metal hole in the panel and via the 2 mounting screws+nuts.
Pin1 is connected to the shell tag.
Pins 2 & 3 connect via a twisted pair to the signal pads on the PCB. Simples.
Sorry guys if this making you nervous, but what happens when the wiper ( or all the relays) opens up?
Then you will in best case ( with the Maya add-on) have full amplification for a couple of mili-seconds, and in worst case (with a regular pot) the opamp will go to full open-loop amplification, and the output will go to one of the supply lines voltage-level.
AndrewT has been the only one mentioning it so far, but this is the Achilles-heel of the system: The wiper must never loose contact!
The positioning of the pot in the feedback line has been done before, in Musical Fidelity A1, and exactly the aging and wear of the pot, which eventually makes the wiper loose contact, has had catastrophic consequences for some A1's.
This means that a relay-based or FET-switched potmeter must use some sort of make-before-break technology, where the wiper always have contact somewhere close to the position it is leaving.
Does the Maya attenuator have a 'fast-on/slow-off" function for activating the relays? ( Hans Polak)
If not - does the Maya attenuator stay relatively silent ( without violent CLICKs) when turning up and down with music on? ( Anyone who buildt it already)
Then you will in best case ( with the Maya add-on) have full amplification for a couple of mili-seconds, and in worst case (with a regular pot) the opamp will go to full open-loop amplification, and the output will go to one of the supply lines voltage-level.
AndrewT has been the only one mentioning it so far, but this is the Achilles-heel of the system: The wiper must never loose contact!
The positioning of the pot in the feedback line has been done before, in Musical Fidelity A1, and exactly the aging and wear of the pot, which eventually makes the wiper loose contact, has had catastrophic consequences for some A1's.
This means that a relay-based or FET-switched potmeter must use some sort of make-before-break technology, where the wiper always have contact somewhere close to the position it is leaving.
Does the Maya attenuator have a 'fast-on/slow-off" function for activating the relays? ( Hans Polak)
If not - does the Maya attenuator stay relatively silent ( without violent CLICKs) when turning up and down with music on? ( Anyone who buildt it already)
There are no schemes using the Maya attenuator with the BPPBP. What you're thinking of is the use of the Maya CPU/display board with Hans' own series resistor switching scheme. He designed it so that are always resistors in that feedback path and which also limit the rang to a useful one.
It has been mentioned way way back in the thread and discussed a little.
It has been mentioned way way back in the thread and discussed a little.
The pot in front of the inverting opamp does not suffer this problem. Basically I adopted the HansPolak attenuation using a switched pot. The opamp sees a fixed feedback resistor and a fixed input resistor
I have a range of -51dB to -0dB in steps of 1.5db and it works smoothly and wonderfully. I used a pair of 5k11 as the fixed feedback resistor. If I remove one of the parallel pair, the attenuation changes by 6dB, i.e the range becomes -45dB to +6dB
I tried the standard linear track pot and found it noisy. Maybe excessive gain.
I took a long time to re-resistor my cheap Chinese stepped pot but got it all working two days ago. Been on test and finding some unexplained problems.
One channel is clearly noisy. I can't identify which opamp is faulty.
Yesterday, I disconnected the internal mains transformer and used a 15-0-15Vac remote transformer to see it that cured or attenuated the noise. It didn't. The regulators got too hot to touch within 5 minutes ! 15Vac is TOO HIGH !
I did hear electrical interference from my neighbour's electric lawn mower. Never heard that before !
I could also hear a ticking noise coming from my DVD player. I have heard this before, but this time it is noticably louder.
I unplugged the remote transformer and found that an attenuated music signal was still passing through the OFF balanced attenuator.
I am going to try to get to the bottom of this today but it happened with lots of combinations of input and output cables. I wonder if it is related to the way the mute relays switch the signal when no mains is present?
The variable output offset problem reported in post450 has gone. The HansPolak circuit cures that.
The post459 problems are three quarters cured:
The excessive gain is gone.
The unbalance between tracks of a linear pot has gone.
The VERY NARROW usable rotation range has gone. The range and steps of the HansPolak style attenuation have have cured this.
The noise remains.
Last edited:
There must be something horribly wrong.
One of the beauties of the BPBP is the high CMRR which should suppress all the kind of funny sounds that you describe.
But also noise should not at all be there.
In the 0dB setting, noise output is ca. 15nV/rtHz, giving max 2uV noise over a 20Khz bandwidth and less with a lower volume setting, really impossible to hear.
Are you sure the Op-Amp's you have are the real one's. Fake's cannot be recognized from the outside, only trust well known suppliers.
Fake ones can easily oscillate, giving in that case all the phenomena that you experience.
The mute relay should give some -60dB attenuation.
I'm using this relay also for a real mute function from my remote control, and what I get is silence.
Let me know if I can help you any further.
Hans
Although this can never happen, feedback would remain active at all times, gain can never become more than 6dB and no DC will ever appear at the output.
But all relays should be closed before going to another volume setting for a period between 10 and 20msec.
So critical changes from -26dB (011111) to -25dB (100000) go like -26/-57/-25. It is impossible to hear this period of 10 to 20 msec, but it fully prevents any clicking noise.
Tibi assured me that is has.
Hans
Last edited:
The PCB is the ready assembled with smds from the Group Buy.
I only added the big components.
I did have to tap into the two (left and right) "net-ties" for the 4th connection for the attenuator. It has a ground in addition to the three pot tappings.
Last edited:
Sounds good Hans.
However still... it is still not optimum regarding the concept of just very short times with max-gain ( even if it is only 6dB because of the safety of the other end of the attenuator-tree)
But if there is a 'fast-on/slow-off' function it could hold the previos settings relay-contacts tied to the chain, while the new settings relay contacs are engaged. This will result in a overall lesser gain in those 10-20msec which is much safer and shouldn't disturb the listening.
So if there is this 'fast-on/slow-off' function incorporated, then that could be the reason that you are not experiencing any clicks or malfunctions.
Hmmm to ground you say? But then the balancing-concept of the BPBP is not fully intact any more.
As I read the description you should connect the 'gnd' end of the attenuator to the op-amp input side GND1-> P1 and GND2 ->P4 ....
Hans, what is the correct way to connect it?
The right channel "net-tie" is connected to the right channel +IN pin.
The left channel "net-tie" is connected to the left channel +IN pin.
Look at the PCB, you can see those connections. They are adjacent to each other, the trace is tiny in length.
The two "net-ties" are separated by a very big plane distance. That is equivalent to a non zero impedance.
The left channel "net-tie" is connected to the left channel +IN pin.
Look at the PCB, you can see those connections. They are adjacent to each other, the trace is tiny in length.
The two "net-ties" are separated by a very big plane distance. That is equivalent to a non zero impedance.
confusing measurements !
The balanced preamp is NOT connected to the Mains power. Only interconnects to the stereo amp.
The DVD player is playing an audio CD but is not connected to any input.
a.) Two outputs to the Stereo amplifier.
a.1) left amp out = 0.1mVac & ~-8.6mVdc
a.2) right amp out = 1.4mVac & ~-28.5mVc
b.) Swap the two interconnects at the stereo amp.
b.1) left amp out = 1.4mVac & ~-8.7mVdc
b.2) right amp out = 0.1mVac & ~-28.5mVc
c.) Disconnect both interconnects at the stereo amp.
c.1) left amp out = 0.1mVac & -8.6mVdc
c.2) right amp out = 0.1mVac & -28.5mVc
That seems to confirm that:
The stereo amp has both channels with roughly the same low noise, even when the inputs are open circuit.
One channel of the balanced pre has one noisy channel and it is consistently noisy when swapped to the other channel of the stereo amp.
d.) Added one zero ohm dummy plug to the input of the bal pre.
d.1 to d.4) Checked all 4 input with dummy plug. No change in noise measurements at amp output.
e.) Swapped the input switch of the bal pre.
e.1 to e.4) Again no change at the amp output.
f.) inserted right channel interconnect (different order from the way I disconnected).
f.1) right bal pre to right stereo amp:
f.1.a) very noticeable clicking noise repeating roughly once per second and pronounced hum. amp out varies continually from 30mVac to 150mVac
f.1.b) left amp out 0.3mVac with open input.
f.2) right bal pre to left stereo amp: quieter clicking noise and slight hum, but clearly audible at listening seat.
f.2.a) left amp out 0.1mVac with open input
f.2.b) right amp out ~19mVac
g.) insert left channel interconnect:
g.1) left bal pre to right stereo amp both amp outs are quiet. 0.1mVac
g.2) left bal pre to left stereo amp, both amp outs are quiet. 0.1mVac
h.) insert both interconnects at both ends: but swapped interconnects compared to tests a. & b.
h.1) right pre to amp out 0.1mVac
h.2) left pre to amp out 1.0mVac
Go back to repeat tests f.
f.1a) touching the disconnected right output from the bal pre changes the hum and noise being sent via the left bal pre output to the stereo amp.
Faulty interconnect?
f.2a) swapped the balanced interconnect:
amp quiet.
I'm going to check the integrity of the suspect cable.
edit:
broken pin3 solder in that suspect cable.
It was a bought in one of a set from Blue Jeans.
I will now check all the others.
The balanced preamp is NOT connected to the Mains power. Only interconnects to the stereo amp.
The DVD player is playing an audio CD but is not connected to any input.
a.) Two outputs to the Stereo amplifier.
a.1) left amp out = 0.1mVac & ~-8.6mVdc
a.2) right amp out = 1.4mVac & ~-28.5mVc
b.) Swap the two interconnects at the stereo amp.
b.1) left amp out = 1.4mVac & ~-8.7mVdc
b.2) right amp out = 0.1mVac & ~-28.5mVc
c.) Disconnect both interconnects at the stereo amp.
c.1) left amp out = 0.1mVac & -8.6mVdc
c.2) right amp out = 0.1mVac & -28.5mVc
That seems to confirm that:
The stereo amp has both channels with roughly the same low noise, even when the inputs are open circuit.
One channel of the balanced pre has one noisy channel and it is consistently noisy when swapped to the other channel of the stereo amp.
d.) Added one zero ohm dummy plug to the input of the bal pre.
d.1 to d.4) Checked all 4 input with dummy plug. No change in noise measurements at amp output.
e.) Swapped the input switch of the bal pre.
e.1 to e.4) Again no change at the amp output.
f.) inserted right channel interconnect (different order from the way I disconnected).
f.1) right bal pre to right stereo amp:
f.1.a) very noticeable clicking noise repeating roughly once per second and pronounced hum. amp out varies continually from 30mVac to 150mVac
f.1.b) left amp out 0.3mVac with open input.
f.2) right bal pre to left stereo amp: quieter clicking noise and slight hum, but clearly audible at listening seat.
f.2.a) left amp out 0.1mVac with open input
f.2.b) right amp out ~19mVac
g.) insert left channel interconnect:
g.1) left bal pre to right stereo amp both amp outs are quiet. 0.1mVac
g.2) left bal pre to left stereo amp, both amp outs are quiet. 0.1mVac
h.) insert both interconnects at both ends: but swapped interconnects compared to tests a. & b.
h.1) right pre to amp out 0.1mVac
h.2) left pre to amp out 1.0mVac
Go back to repeat tests f.
f.1a) touching the disconnected right output from the bal pre changes the hum and noise being sent via the left bal pre output to the stereo amp.
Faulty interconnect?
f.2a) swapped the balanced interconnect:
amp quiet.
I'm going to check the integrity of the suspect cable.
edit:
broken pin3 solder in that suspect cable.
It was a bought in one of a set from Blue Jeans.
I will now check all the others.
Last edited:
So it seems that you finally found a cause for some or even for all of your preamp trouble ?
That would be great news.
Hans
Much better.
Both good cables connected to an OFF bal pre and the amp out is 0.1mVac
Disconnect any one and the noise increases to 0.3mV to 0.4mV in the still connected channel.
Swap the leads around and the connected channel increases to 1.7mVac from the left output and the disconnected channel reads 0.1mVac
The left out of the bal pre is still giving a higher noise in it's connected channel.
The ticking noise although very quiet is still there when one channel is connected.
More tests to follow with DVD sending signal.
Both good cables connected to an OFF bal pre and the amp out is 0.1mVac
Disconnect any one and the noise increases to 0.3mV to 0.4mV in the still connected channel.
Swap the leads around and the connected channel increases to 1.7mVac from the left output and the disconnected channel reads 0.1mVac
The left out of the bal pre is still giving a higher noise in it's connected channel.
The ticking noise although very quiet is still there when one channel is connected.
More tests to follow with DVD sending signal.
I think you missed the point. It is not a short period of max gain, but a period of minimum gain instead at -57dB.
So all relays in the on position remain in that position, and all going to a off position do this a bit later.
Have another look at the circuit diagram and you will understand that relays "in between" on and off do no harm at all.
This is exactly what I described, fast-on/slow-off, the safest method there is.
No the feedback cancellation for a pot is only partly applicable (4k/2K with constant 6dB gain) but in front is a divider which is equiped with resistors not behaving like a potmeter track.
The exact way to connect is here:
Hans
Non Linearity might not be the real issue
I think there may be a few assumptions that are not as important as some people think. I am not criticizing the math that Hans has provided.
Hans and perhaps Bruno is assuming that the non linear aspect of the pot is uniform and will be cancelled with the circuit topology as suggested in Bruno's circuit. If the pot actually exhibits the behavior as suggested by Han's equation, then we should not expect it to be frequency dependent.
Bruno's measurement of a log pot shows a rise in low frequency distortion. This is a bit analogous to adding a resistor on the end of the potentiometer.
I think the phenomena that we are observing is caused by thermal modulation. If we split the potentiometer into small equal lengths of ink, then the IR drop across the same length will be uniform in a linear taper. The total resistance will change as a function of the voltage across the potentiometer, but the ratio will stay constant. This is why the linear pot had better low frequency distortion. The log pot would have hot spots since the IR drop would be different for the same segment length. At low frequencies this temperature is modulated, but as frequency increases the effect will stabilize. It will slightly change the set point of the pot, but not the distortion.
This is consistent with Bruno's measurement.
Now certainly adding a resistor on the ends of the pot will exhibit the same problem because the resistance will change a little for the same reasons and the thermal relationship will likely be different.
This same problem can occur in a logarithmic attenuator wired in the same topology as the potentiometer. If the elements are all the same, then we would have the same thermal cancellation as the linear pot. You would just have a lot of elements and probably lots of missing switching taps if you were roughly fitting logarithmic level changes.
Now if you use parts with much less susceptibility to thermal modulation, for example a larger wattage resistor of the same type or a better type, this distortion will be minimized.
I bring this up because I have a real problem with a design that essentially goes open loop when the pot is at the end. Trust me, this is a distortion a deaf person could hear (or perhaps see when the tweeter bursts into flames)
Now whether the thermal distortion is clearly audible or not, I don't know. I haven't done the critical listening tests for this situation.
I should point out that I have no disagreement with the rest of the circuit and its ideas. It is clearly well reasoned.
There is one more thing to consider. If you please two resistors of equal value at each end and a linear pot in the middle, then the center position cancels the distortion. This might be good for small adjustments like speaker sensitivity where the center is defined as normal and the deviation are tweaks. The distortion will increase as you go to the end points, but you can keep all the levels in reasonable boundaries.
Al Clark
Danville Signal
I think there may be a few assumptions that are not as important as some people think. I am not criticizing the math that Hans has provided.
Hans and perhaps Bruno is assuming that the non linear aspect of the pot is uniform and will be cancelled with the circuit topology as suggested in Bruno's circuit. If the pot actually exhibits the behavior as suggested by Han's equation, then we should not expect it to be frequency dependent.
Bruno's measurement of a log pot shows a rise in low frequency distortion. This is a bit analogous to adding a resistor on the end of the potentiometer.
I think the phenomena that we are observing is caused by thermal modulation. If we split the potentiometer into small equal lengths of ink, then the IR drop across the same length will be uniform in a linear taper. The total resistance will change as a function of the voltage across the potentiometer, but the ratio will stay constant. This is why the linear pot had better low frequency distortion. The log pot would have hot spots since the IR drop would be different for the same segment length. At low frequencies this temperature is modulated, but as frequency increases the effect will stabilize. It will slightly change the set point of the pot, but not the distortion.
This is consistent with Bruno's measurement.
Now certainly adding a resistor on the ends of the pot will exhibit the same problem because the resistance will change a little for the same reasons and the thermal relationship will likely be different.
This same problem can occur in a logarithmic attenuator wired in the same topology as the potentiometer. If the elements are all the same, then we would have the same thermal cancellation as the linear pot. You would just have a lot of elements and probably lots of missing switching taps if you were roughly fitting logarithmic level changes.
Now if you use parts with much less susceptibility to thermal modulation, for example a larger wattage resistor of the same type or a better type, this distortion will be minimized.
I bring this up because I have a real problem with a design that essentially goes open loop when the pot is at the end. Trust me, this is a distortion a deaf person could hear (or perhaps see when the tweeter bursts into flames)
Now whether the thermal distortion is clearly audible or not, I don't know. I haven't done the critical listening tests for this situation.
I should point out that I have no disagreement with the rest of the circuit and its ideas. It is clearly well reasoned.
There is one more thing to consider. If you please two resistors of equal value at each end and a linear pot in the middle, then the center position cancels the distortion. This might be good for small adjustments like speaker sensitivity where the center is defined as normal and the deviation are tweaks. The distortion will increase as you go to the end points, but you can keep all the levels in reasonable boundaries.
Al Clark
Danville Signal
Just don't move the pot it's full travel? You can easily prevent this from happening with via physical means. Hans' Vol Control Board gets around this anyway so the point is kind of moot.
Remember, this circuit is ONLY a proof of concept, concept not limited to the pot used in this position, mainly proof of concept regarding his concept of ground and differential circuit design.
It was never intended that the circuit is put in use as a permanent piece of equipment! But, we have CHOSEN to do so.
. Those doing so must live with the fact that it is a proof of concept, or otherwise mitigate any long term problems e.g Hans' contribution, that he made many years ago when the article was published in Linear Audio.
Remember, this circuit is ONLY a proof of concept, concept not limited to the pot used in this position, mainly proof of concept regarding his concept of ground and differential circuit design.
It was never intended that the circuit is put in use as a permanent piece of equipment! But, we have CHOSEN to do so.
. Those doing so must live with the fact that it is a proof of concept, or otherwise mitigate any long term problems e.g Hans' contribution, that he made many years ago when the article was published in Linear Audio.