I guess you don't believe that the 18db result is real if each resistor in the pair is nearly exactly equal 509/509 and 179/179 i.e. gain very close to 1. There is no >10% difference to get CMRR 18db(if not including your DAC output impedance). Why don't you test CMRR as everyone else does?
Last edited:
Well I measured one leg at the time and did not get gain of 1. More like 70% difference in gain between legs.
Normally two signals with the same amplitude and phase, can be considered as one V1
Normally two signals with the same amplitude and phase, can be considered as one V1
I think there is a problem with the circuit shown, if the source drives the two signals separately and the source impedance is not 0 ohm. If both legs of the signal have an impedance of around 110 ohm, you will get a CMRR of 18dB.
So the circuit is not fully balanced, since the input impedances are different.
So the circuit is not fully balanced, since the input impedances are different.
At least it is very important that it is documented, then the user can adapt.
But do you have gain of 1 on each leg?
But do you have gain of 1 on each leg?
for sure, 180/180 = 1 and 510/510 = 1, isn't it? And differential impedance is also matched(almost) 😉
CMRR test is not what you did but what I did as shown in the pic above.
CMRR test is not what you did but what I did as shown in the pic above.
And is the resistor values the same as in my picture?
That is 180 or 510. I wonder if some have the wrong placement on my board
That is 180 or 510. I wonder if some have the wrong placement on my board
I don't remember if an SMT P&P machine ever mixed up parts and after that machine vision inspection missed that but you can compare it with my LPF pic if you like.
https://e1dashz.wixsite.com/index/cosmos-apu
https://e1dashz.wixsite.com/index/cosmos-apu
OK, see it's the same as mine.
Don't know how to get further.
You have 0 dB gain on both legs.
I have -6dB and -14dB
I can live with it. I will report back if any faults on my side. But the switch opening and closing are pretty foolproof🙂
Don't know how to get further.
You have 0 dB gain on both legs.
I have -6dB and -14dB
I can live with it. I will report back if any faults on my side. But the switch opening and closing are pretty foolproof🙂
Yes, the output Zout of source is the problem.
Guess Zin of the unbal leg of 180 R is way to low for my source.
Did not see it on the LPF part as it is the same on both legs and almost 10x the R.
Guess Zin of the unbal leg of 180 R is way to low for my source.
Did not see it on the LPF part as it is the same on both legs and almost 10x the R.
Made a ltspice simulation of the unbal circuit with source resistors. But still can’t get my grasp around the big difference between the gains caused by the source resistors.
Ltspice show much less attunation than my worst case measurements.
My understandin is of input Z
2R on on 180 leg
2/3R on 510 leg
Gain 0.5 by adding source R
+2R source on 180 leg
+R source on 510 leg
Probably still overlooking something
Ltspice show much less attunation than my worst case measurements.
My understandin is of input Z
2R on on 180 leg
2/3R on 510 leg
Gain 0.5 by adding source R
+2R source on 180 leg
+R source on 510 leg
Probably still overlooking something
picture #861
for a balanced input, R1=R3 and R2=R4. if not, the non-zero output impedance of the source will mess up the cmrr (even if it is the same on both legs)
for a balanced input, R1=R3 and R2=R4. if not, the non-zero output impedance of the source will mess up the cmrr (even if it is the same on both legs)
Last edited:
This trick only shall ever be used inside a circuit but never as an input stage, as it violates the #1 requirement for balanced input stages, equal impedances on both legs. In others words, CMRR shall not suffer as long as source impedance is balanced, regardless of absolute value. The 3:1 resistor scaling clearly violates that. And it only balances currents when the input signal is balanced. See http://www.douglas-self.com/ampins/balanced/balanced.htm#5
If one wants equal impedances and equal currents (instead of the 3:1 ratio) one must use an FDA (integrated or emulated).
correct, and we are talking about balance2unbalance stage i.e. the input signal is balanced, and input currents are equal.