Hello all,
I moved here for the moment :
https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7795028
It will be again couple of weeks and than I can continue running measurements with test LP on TT as signal source....
I moved here for the moment :
https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7795028
It will be again couple of weeks and than I can continue running measurements with test LP on TT as signal source....
Sorry, one more thing, I thing last proposal for first gain stage, as Hans posted in #362 could be very promising. To start NE5534 could be used for U1-2 and TL072 for U3-4, all cheap and cheerful. When I manage Ill make PCB, if someone else has time and will to draw it, superb. This will be too complex to hard wire.
For next stage, RIAA filter, I kind of already have it fixed for large precision polystyrene capacitors. This is bit against my philosophy that everything should be made of components from the shop, and easily available... But with 1% precision, negligible inductance and excellent stability, I cant resists them.... Even they do occupy a lot of space and could pick noise..
This is what I have in mind:
For next stage, RIAA filter, I kind of already have it fixed for large precision polystyrene capacitors. This is bit against my philosophy that everything should be made of components from the shop, and easily available... But with 1% precision, negligible inductance and excellent stability, I cant resists them.... Even they do occupy a lot of space and could pick noise..
This is what I have in mind:
Very nice Calvin. Now if we add a step up transformer to provide galvanic isolation and improve input noise, run it off higher voltage rails - extra headroom - it would be just about perfect 😇
Hi,
thanks mate 😉
Apart from the galvanic isolation -why should a SUT matter or make a difference at this location? - the circuit is just about perfect as is.
😆
The optimal noise source resistance of these INAs is around 200R.
A SUT really won´t improve matters here. Pickups down to 0.15mV are amplified wo. relevant audible noise.
+-15V supply lines allow for roundabout +20dBV, and which is more than sufficient headroom ... with a nominal 300mV (~-10dBV) output level this translates to close to +30dB of headroom.
Higher supply lines rather only generate more heat power losses.
jauu
Calvin
thanks mate 😉
Apart from the galvanic isolation -why should a SUT matter or make a difference at this location? - the circuit is just about perfect as is.
😆
The optimal noise source resistance of these INAs is around 200R.
A SUT really won´t improve matters here. Pickups down to 0.15mV are amplified wo. relevant audible noise.
+-15V supply lines allow for roundabout +20dBV, and which is more than sufficient headroom ... with a nominal 300mV (~-10dBV) output level this translates to close to +30dB of headroom.
Higher supply lines rather only generate more heat power losses.
jauu
Calvin
Hello again Calvin,
Glad to see your reply. One of the interesting things about a forum like this is that from time to time certain posts cause one to reassess preconceived ideas. So lets have a rethink; let’s start from first principles and do some back of the envelope, first order, calculations.
For the INA 103 the current noise density, j, varies with frequency. From the data sheet take j = 5 pA Per root Hz, which is the value near LF where RIAA boost will be applied. Take voltage noise to be 1 nV per root Hz. Let us take a typical MC cartridge with resistance of 10 ohms. The actual output is unimportant as were just comparing the noise in the two cases, with and without transformer.
Without the transformer, the system noise will be:
1. 1 nV /rt Hz from the op amp.
2. 10 Ohm resistance from the coil of the cartridge—> 0.4 nV/ rt Hz
3 Current noise = j*R = 5 x E-12 * 10 , negligible.
Net e = sq rt(1^2 + 0.4^2) = 1.1 nV/ rt Hz
Not really relevant to the current discussion, but the INA 103 is significantly noisier than discrete ZTX based phono stages presented here, or the older Paradise.
Now, let’s look at the transformer coupled stage. It’s very hard to get information on primary and secondary resistances of transformers from mainstream manufacturers, so let’s consider a transformer for which data is available, e.g. Jensen 346. With primaries connected in series, and secondaries also connected in series, the manufacturer’s data sheet gives a DC resistance of 6 Ohms for the primary and 174 ohms for the secondary with a step up ratio of 4 x.
Referred to the primary, we have:
1. 1/nV /Hz /4 for the op amp = 0.25 nV/Hz
2. (10+6) ohms total resistance in the primary, noise contribution = 0. 5 nV/Hz
3. The effective resistance in the secondary = (10+6)*4^2 + 174 = 430 Ohms.
Noise in the secondary due to this resistance = 430 * 5E-12 = 2.2 nV/Hz.
Now reflect this back to the primary, —> 2.2/4 = 0.53 nV/Hz.
Total noise density= sq rt(0.53^2 + 05 ^2. + 0.25^2) = 0.8 nV/Hz
Note that this is lower than the direct connection without SUT, which is what I was getting at in my original post.
Interestingly, if you repeat this compilation with 12 times step up on the same Jensen SUT, the noise increases. There is between 4x and 12 x an optimum turns ratio for minimum noise. It should be pretty straightforward to write out an algebraic expression in terms of resistances, noise densities, the turns ratio, etc., differentiate with respect to it and find the optimum.
The culprit here is, of course, the current noise in the INA 103. If the current noise were to be reduced by adding a low noise FET buffer between the transformer and the INA 103, we could use higher turns ratios and get lower system noise.
Naturally, this is a very simplified model and does not take into account other noise sources within the transformer, etc.
Regarding heating due to higher supply voltages, I honestly don’t think this is an issue. The difference in chip temperatures when the voltage raised is probability less than the difference in temperatures between Berlin and Phoenix, Arizona.😄. I mean people sweat blood to get 1 or 2 dB lower noise, so why not take advantage of higher supply voltage to get more headroom when you can.
Enough for now, it’s late and I had a looong weekend 😊.
Comments and corrections are welcome.
Glad to see your reply. One of the interesting things about a forum like this is that from time to time certain posts cause one to reassess preconceived ideas. So lets have a rethink; let’s start from first principles and do some back of the envelope, first order, calculations.
For the INA 103 the current noise density, j, varies with frequency. From the data sheet take j = 5 pA Per root Hz, which is the value near LF where RIAA boost will be applied. Take voltage noise to be 1 nV per root Hz. Let us take a typical MC cartridge with resistance of 10 ohms. The actual output is unimportant as were just comparing the noise in the two cases, with and without transformer.
Without the transformer, the system noise will be:
1. 1 nV /rt Hz from the op amp.
2. 10 Ohm resistance from the coil of the cartridge—> 0.4 nV/ rt Hz
3 Current noise = j*R = 5 x E-12 * 10 , negligible.
Net e = sq rt(1^2 + 0.4^2) = 1.1 nV/ rt Hz
Not really relevant to the current discussion, but the INA 103 is significantly noisier than discrete ZTX based phono stages presented here, or the older Paradise.
Now, let’s look at the transformer coupled stage. It’s very hard to get information on primary and secondary resistances of transformers from mainstream manufacturers, so let’s consider a transformer for which data is available, e.g. Jensen 346. With primaries connected in series, and secondaries also connected in series, the manufacturer’s data sheet gives a DC resistance of 6 Ohms for the primary and 174 ohms for the secondary with a step up ratio of 4 x.
Referred to the primary, we have:
1. 1/nV /Hz /4 for the op amp = 0.25 nV/Hz
2. (10+6) ohms total resistance in the primary, noise contribution = 0. 5 nV/Hz
3. The effective resistance in the secondary = (10+6)*4^2 + 174 = 430 Ohms.
Noise in the secondary due to this resistance = 430 * 5E-12 = 2.2 nV/Hz.
Now reflect this back to the primary, —> 2.2/4 = 0.53 nV/Hz.
Total noise density= sq rt(0.53^2 + 05 ^2. + 0.25^2) = 0.8 nV/Hz
Note that this is lower than the direct connection without SUT, which is what I was getting at in my original post.
Interestingly, if you repeat this compilation with 12 times step up on the same Jensen SUT, the noise increases. There is between 4x and 12 x an optimum turns ratio for minimum noise. It should be pretty straightforward to write out an algebraic expression in terms of resistances, noise densities, the turns ratio, etc., differentiate with respect to it and find the optimum.
The culprit here is, of course, the current noise in the INA 103. If the current noise were to be reduced by adding a low noise FET buffer between the transformer and the INA 103, we could use higher turns ratios and get lower system noise.
Naturally, this is a very simplified model and does not take into account other noise sources within the transformer, etc.
Regarding heating due to higher supply voltages, I honestly don’t think this is an issue. The difference in chip temperatures when the voltage raised is probability less than the difference in temperatures between Berlin and Phoenix, Arizona.😄. I mean people sweat blood to get 1 or 2 dB lower noise, so why not take advantage of higher supply voltage to get more headroom when you can.
Enough for now, it’s late and I had a looong weekend 😊.
Comments and corrections are welcome.
Last edited:
P.S:- for 12 x turns ratio, the primaries are connected in parallel, giving a primary assistance Of approx. 0.7 Ohm, The secondary remains unchanged.
My way of calculating is a bit different.
1nV/rtHz for the INA equals the noise of a 60R resistor.
Add this to the 174R of the secondary winding, giving an equivalent resistance of 60+174 = 234R.
Bring this to the primary by dividing this value by the (Turns Ratio)^2 gives 234/(4)^2 = 14.6R.
Now add this to the 10R from the Cart and the 6R from the primary winding and you get 14.6+10+6 = 30.6R
Now the equivalent input noise is sqrt(30.6/60) = 0.71nV/rtHz.
Hans
Hi,
Everyone in family was sick, one after another, so bit less time, but it is getting better. At least managed to repurpose desk completely for TT use, connected original preamp in pieces and listen to some LP's . Test LP is still traveling, as soon as it is there I can post some hopefully meaningful measurements
But I have one important question:
Finally dared to measure R of MC cart coils. Cant measure it precisely due to contact + cable resistances, but what is consistent is that one channel is about 3 Ohm less than the other. Given median R of 40Ohm, that is 7.5% difference 🤔
If trans-impedance amp (where RG is Rcart) is used on this one, I will get 7.5% gain difference between channels, hmmmmm, or did I get it wrong?
Looking forward to test LP so I can also compare 2 channels V output too..
Did anyone observe such difference between L and R coil?
Everyone in family was sick, one after another, so bit less time, but it is getting better. At least managed to repurpose desk completely for TT use, connected original preamp in pieces and listen to some LP's . Test LP is still traveling, as soon as it is there I can post some hopefully meaningful measurements
But I have one important question:
Finally dared to measure R of MC cart coils. Cant measure it precisely due to contact + cable resistances, but what is consistent is that one channel is about 3 Ohm less than the other. Given median R of 40Ohm, that is 7.5% difference 🤔
If trans-impedance amp (where RG is Rcart) is used on this one, I will get 7.5% gain difference between channels, hmmmmm, or did I get it wrong?
Looking forward to test LP so I can also compare 2 channels V output too..
Did anyone observe such difference between L and R coil?
So far only managed to make shots of
1. cart directly plugged in sound card, bottom trace or green
2. and my old preamp powered up and connected to cart, upper trace or blue
Both without signal, arm in the rest but new motor running.
Picture is speaking for itself i think. Very useful but really a lot of noise form preamp, however in use I don't notice it...
There is also ugly 50Hz peak, at moment connected to regulated transformer based PS and without shielding, this I feel comfortable to manage in later steps
Last thing to observe is there is no motor noise visible (it runs on 30hz now) which is good.
1. cart directly plugged in sound card, bottom trace or green
2. and my old preamp powered up and connected to cart, upper trace or blue
Both without signal, arm in the rest but new motor running.
Picture is speaking for itself i think. Very useful but really a lot of noise form preamp, however in use I don't notice it...
There is also ugly 50Hz peak, at moment connected to regulated transformer based PS and without shielding, this I feel comfortable to manage in later steps
Last thing to observe is there is no motor noise visible (it runs on 30hz now) which is good.
Hello Hans,
It is always a pleasure to have your input.
In your calculations I see resistances and their voltage noise, but do not see any consideration of the current noise of the INA 103. I am unclear on how that is taken into account in your calculations? As the turn ratio increases, the effective resistance reflected from the primary to the secondary will increase by N^2, and Noise due to the INA 103 current noise will correspondingly increase. That noise will be reflected back to the primary, no? That is why the 12x turns ratio is noisier than 4 x ratio. I do believe a current noise term for the secondary should be included in the calculations.
Best.
Last edited:
Yes, I omitted the current noise assuming it could be neglected, but you are right.
2pA/rtHz times 174R gives 0.35nV/rthz, equivalent to another 7.4R resistance.
To do the calculation again (174+60+7.4)/16 = 15.1R.
sqrt((15.1+6+10)/60) = 0.72nV/rtHz
So adding the current noise makes only a very small difference of 0.01nV/rtHz.
Hans
So the difference in our calculations is that you are not including the resistance (10+6) * 4^2 = 256 R in your current noise calculations? For the secondary, I am using 430 Ohms, you are using 174 Ohms to calculate current noise.
Hi,
just a remark regarding the noise calculations .... either using a transformer or the INA directly - is the small difference in noise figures practically relevant in comparison to the maximum SND that Vinyl allows for?
The INA input stage circuits are at least 10dB better than that said limit.
So, if it doesn´t make a difference in praxis, I wonder what could be a real benefit in adding a SUT here, regarding the amount of linear and nonlinear stray parameters it adds to the equations?
jauu
Calvin
just a remark regarding the noise calculations .... either using a transformer or the INA directly - is the small difference in noise figures practically relevant in comparison to the maximum SND that Vinyl allows for?
The INA input stage circuits are at least 10dB better than that said limit.
So, if it doesn´t make a difference in praxis, I wonder what could be a real benefit in adding a SUT here, regarding the amount of linear and nonlinear stray parameters it adds to the equations?
jauu
Calvin
Something I am wondering about Hans. I am in the process of constructing a fully balanced "current mode" MC preamplifier that uses the SSM2212 for the differential input pair, this in lieu of using a pair of ZTX851's.
The thermal tracking proved problematic even with DC servo feedback using the ZTX devices (In contrast using the SSM2212 questions if the DC servo is even needed). The input network looks as per the following with about 1.8mA per leg. The resistors are 0.1 % to support fine tuning.
https://www.analog.com/media/en/technical-documentation/data-sheets/SSM2212.pdf
Besides the noise "seeming" good on an oscilloscope... what would you expect the noise to be in this "current mode" application, given a Denon 103R with 0.3mV and coil resistance of 14 Ohms? It seems the noise would be dominated by the SSM2212, though it isn't clear when operating in current mode. What significance does the 2x emitter resistances have across the coil? For example what if the coil resistance was 0 Ohms?
Hi Hierfi,
Something similar I tested in https://www.diyaudio.com/community/...ono-preamplifier-thoughts.414816/post-7751410
It was SSM2220 which is PNP, but believe very similar to SSM2212 otherwise.
Something similar I tested in https://www.diyaudio.com/community/...ono-preamplifier-thoughts.414816/post-7751410
It was SSM2220 which is PNP, but believe very similar to SSM2212 otherwise.
Since test LP is still not in, and I had a bit of time, I run some sanity checks with old preamp and signal from soundcard, reduced as I did before;
The first stage is actually double, so I can check it as balanced and single ended output: both results as expected:
Balanced; signal taken between A and B:
Then single ended output; signal taken between B and ground. Exactly as expected, slightly worse SNR and THD, but all good.
Reminder, this is with 33R resistor at input, not with cable through the arm and than cart:
The first stage is actually double, so I can check it as balanced and single ended output: both results as expected:
Balanced; signal taken between A and B:
Then single ended output; signal taken between B and ground. Exactly as expected, slightly worse SNR and THD, but all good.
Reminder, this is with 33R resistor at input, not with cable through the arm and than cart:
INA 103 ‘sees’ the reflected primary 256 Ohm resistance - it is in series with 174 Ohms - so I expect it should be included noise calculations.
Will give it second look when I have time……