TLC4502 as servo -- a bit lower delta between the emitters -- lower hundreds of uV -- seems to work OK -- just using 5V and the TLC4502 is running off the 5V supply. Will check the noise with these devices tomorrow.
I think you will need to keep the max offset to 10-20uV. I have no formula for why, but it seems DC through the coil should be kept to a minimum. Looking forward to your noise measurements.
Not sure - I've heard for and against. We need @billshurv and @luckythedog dog here to answer these questions.
I think it should not be more than a few uA - say <10% of the signal current.
I think it should not be more than a few uA - say <10% of the signal current.
No evidence that DC through the coil moves it all all. The generator is not efficient enough to act as a motor. Would love someone to prove me wrong oneday...
Otherwise said, cartridges are not reciprocal motors. Current through cartridge is one of those ingredients that can make or break a design, based on nothing but hearsay, myths and legends from the High End Audio folklore.
Which doesn't mean it's not worth taking it as a challenge. Assuming an input cap is unacceptable (another fetish) my preference goes toward a floating design, but I don't see why servos won't do too. One interesting property is that input current cancellation (by whatever method) doesn't mean the noise current is zero. Noise are uncorrelated sources, so they still add up, even if the net DC current through cartridge is zero.
Which doesn't mean it's not worth taking it as a challenge. Assuming an input cap is unacceptable (another fetish) my preference goes toward a floating design, but I don't see why servos won't do too. One interesting property is that input current cancellation (by whatever method) doesn't mean the noise current is zero. Noise are uncorrelated sources, so they still add up, even if the net DC current through cartridge is zero.
Last edited:
I am going to set about measuring the noise. I established a setup which uses D-Cells for the voltage source. This necessitated changing some resistor values for Vb such that~10 mA collector current flows. Would like to base-line the measurements using 61 Ohm and 380 Ohm resistors from emitter to emitter, 1.0 and 2.5 nV/Rt Hz respectively. Please feel free to toss a brick at my head (the switches are just jumpers on the solderless breadboard):
Here’s a brick; measuring those levels of noise on a solderless breadboard will fail miserably. You breath on it, noise are all over the place.
well, yes and no. The solderless breadboard goes into a large cookie tin which cookie tin, in and of itself, is connected to the chassis of the AP a la Van der Ziel. The amplifier is connected to the AP by an 8 inch run of RG174/U. Another brick please.
Let’s try it and see. I am pleased that Jack has bread boarded this thing. The 1/f noise is likely to be higher, but for bread board up like this it’s ok. If you’ve got it right, the 1 kHz spot noise will be around 500 pV/rt Hz. I expect on the final PCB based board we can get it reliably to around sub 350 pV/rt Hz and with low 1/f corner (under 100 Hz)
Where did you get your 1/f noise at under 100Hz.
Various measurements with the ZTX851 have shown that 1/f noise corner is in the order of only a few Hz.
Hans
NB you will have to max out the number of averages to get a good reading. I use the full 50 on the QA401
This is a bread boarded design. The 1/f will be thermally induced. I expect on a PCB and gousedvit will be at spec and then a few Hz.
If you have breadboarded and then built on PCB thermocouple amplifiers you will recognize the same issue
If you have breadboarded and then built on PCB thermocouple amplifiers you will recognize the same issue
Agree on the thermal issues -- I saw this first hand as I built Groner's 1000x LNA. When I put it on PCB the performance came close to what was published in Linear Audio. Putting in the cookie tin allows everything to stabilize and knocks down the environmental noise.
The issue with the circuit I drew last night -- the gain will change in all configurations.
Tex Instruments application note on measuring balanced amplifier noise AC couples the input to ground with electrolytic capacitors. -- these have their own noise and settling time issues -- and gain is just ~ Rc/r'e
I derived some of the MC Balanced LNA resistance values for Ic of 10mA with ~6V supply. Will build it dead-bug style when the values are normalized.
The issue with the circuit I drew last night -- the gain will change in all configurations.
Tex Instruments application note on measuring balanced amplifier noise AC couples the input to ground with electrolytic capacitors. -- these have their own noise and settling time issues -- and gain is just ~ Rc/r'e
I derived some of the MC Balanced LNA resistance values for Ic of 10mA with ~6V supply. Will build it dead-bug style when the values are normalized.
Hans
But unlike input caps, servos in discrete input designs have a second function which is to keep the DC collector currents the same. Vbe differences that would cause this are corrected, thereby preventing IM distortion to raise.
For this reason the emitter resistors in Bonsai’s design have to be tightly matched.
Very true, and a needle doesn’t have more problems following the track as is sometimes mentioned when permanent DC current is flowing. The only thing to prevent is to damage the coils from too much current at start up. A few hundred uA are pretty harmless but when it goes in the multi mA it might become risky.
But unlike input caps, servos in discrete input designs have a second function which is to keep the DC collector currents the same. Vbe differences that would cause this are corrected, thereby preventing IM distortion to raise.
For this reason the emitter resistors in Bonsai’s design have to be tightly matched.
I don’t think there is much difference in the number of parts, but doesn’t the word cheap also have the meaning of having a lower quality, certainly not what you meant.
Both designs are made with TLC.