Thanks very much 🙂 , but there is no need. I will build my own.
And how is your I/V converter? already finished ?
And how is your I/V converter? already finished ?
Oh, I have the PCB's drawn up, but I think I want to avoid SMD's so I have to change it to different parts. I'm testing the combination of an LME49710 and a BUF634 instead of a LME49990 and a LME49600. I read that the LME49990 was difficult to prevent from oscillating, for one thing, and another is that while I can do SMD's, it is very time consuming for me to solder them. Thanks for asking.
LME49710 it seems to be a nice opamp and is cheap, i will buy some components , i will include LME49710 in the list.
Yesterday i have spend some time measuring distortion in some opamp´s, and discrete buffers, i have collected some data, this week i will open a new thread about this.
one opamp that really suprise me was the old ne5534, very low noise and low distortion (0.00002%) even with 510 ohms load.
There are some videos in youtube with some techniques for soldering smd. i manage to solder them with relative easy.
Yesterday i have spend some time measuring distortion in some opamp´s, and discrete buffers, i have collected some data, this week i will open a new thread about this.
one opamp that really suprise me was the old ne5534, very low noise and low distortion (0.00002%) even with 510 ohms load.
There are some videos in youtube with some techniques for soldering smd. i manage to solder them with relative easy.
Yes, the NE5534 is a great op amp, and very inexpensive. I'll take a look at it again. The LME49710 has a much higher slew rate than the NE5534. The slew rate of the latter is a function of the compensation capacitor, varying from about 12V/usec to 4V/usec for a 42pF Cc. The GBWP is much higher for the LME49710 also (55Mhz vs. 10Mhz).
I use a clip to hold the SMD part in place, then solder it. For multi-leg devices, I lay the solder across the legs and give a quick touch with my iron. Then I clean up any bridges with solder wick. I can solder THD's much faster though.
Last edited:
sorry for the delay Dirk ,
Probably the LME49710 is a better opamp than ne5534 in every aspects,
I´m just telling that the ne5534 is better than what I thought, but normally i use opa134.
Probably the LME49710 is a better opamp than ne5534 in every aspects,
I´m just telling that the ne5534 is better than what I thought, but normally i use opa134.
I guess the OPA134 is one of the opamps that TI inherited when they bought Burr-Brown.
Anyway, I've redesigned the "zero field I/V converter" board again to accept through hole devices. I'm scared of the LME49990. GBWP is something like 110MHz!
If you don´t feel at ease in designing SMD pcb´s then is better to use a more slow device, at high frequency even the smallest parasitic inductance matters, so in my oppinion you should use the lme49710,
In my personal oppinion i think you should try another type of I/V converter, transformers distortion is low order harmonics and decrease as the frequency increases, i don´t think is necessary to use this zero field I/V converter . why don´t you open a thread about your discrete I/V design, i will try to help you achieve your goals.
In my personal oppinion i think you should try another type of I/V converter, transformers distortion is low order harmonics and decrease as the frequency increases, i don´t think is necessary to use this zero field I/V converter . why don´t you open a thread about your discrete I/V design, i will try to help you achieve your goals.
I'm not interested in pursuing a discrete design because of what Mr. Curl said in the thread about designing a discrete opamp. After I realized that my chances of success with a discrete design were much lower than with IC's, I abandoned the idea.
I'm not completely uncomfortable with designing a SMD board, but I am not an RF engineer so my chances for success with something as wideband as the LME49990 are lower. It's not rocket science to put the passive parts as close to the opamp's pins as possible after all. Given the fact that the manufacturer recommends multiple bypass capacitors for each power pin in order to avoid oscillation, I have to stop and wonder about this device and if the small improvement in performance is worth the effort to make it stable.
I'm much more comfortable with the robustness of a transformer coupling, and the zero field input technique reduces distortion dramatically, so that's my choice.
I'm not completely uncomfortable with designing a SMD board, but I am not an RF engineer so my chances for success with something as wideband as the LME49990 are lower. It's not rocket science to put the passive parts as close to the opamp's pins as possible after all. Given the fact that the manufacturer recommends multiple bypass capacitors for each power pin in order to avoid oscillation, I have to stop and wonder about this device and if the small improvement in performance is worth the effort to make it stable.
I'm much more comfortable with the robustness of a transformer coupling, and the zero field input technique reduces distortion dramatically, so that's my choice.
Dirk, for those of us who have no clue what a "zero field I/V converter" is can you please elaborate?
Is this just a name you have given a topology commonly known under another nomenclature or something unique that you have come up with. I ask because I am relatively conversant with high performance IV design from my days in semi con ATE and have never heard the term. Google turned up nothing that didn't point at this thread.
Thanks! 😀
Is this just a name you have given a topology commonly known under another nomenclature or something unique that you have come up with. I ask because I am relatively conversant with high performance IV design from my days in semi con ATE and have never heard the term. Google turned up nothing that didn't point at this thread.
Thanks! 😀
I adapted the zero field input transformer circuit for use as an I/V converter for a current output DAC. The ZF circuit was patented a long time ago and is sometimes used in pro gear. I reasoned that if I could find a transformer with extremely low DC winding resistance, that I could then use it in this circuit and obtain a lot of benefits including low input impedance, low distortion, frequency response extending to less than 10Hz, galvanic isolation from digital circuits, some RF filtering, and cancellation of second harmonic distortion and any other common mode distortion. So, it's not perfect and I have not built it in real life yet, but I think it will be good enough for me. I have boards drawn up, I just need to pay for them.
See Lundahl's discussion of the ZF circuit in their datasheets for the LL6404 and LL7101 transformers. I am using their LL1517, which has very low winding resistance. I did not use resistors R1 and R2 in order to minimize the input impedance.
Dirk,
Interesting design, thanks for posting it. Not traditionally a technique used for I/V conversion, obviously intended to leverage transformer performance by loading into virtual earth allowing very low inductance to provide extended LF response. I assume if the input resistors are made suitably small (or deleted as you have done) and a dc path were provided where needed it would work.
I use Sowter 9545 in my I/V converter circuit with a 5842 for some additional gain and a lower output Z.
Interesting design, thanks for posting it. Not traditionally a technique used for I/V conversion, obviously intended to leverage transformer performance by loading into virtual earth allowing very low inductance to provide extended LF response. I assume if the input resistors are made suitably small (or deleted as you have done) and a dc path were provided where needed it would work.
I use Sowter 9545 in my I/V converter circuit with a 5842 for some additional gain and a lower output Z.
Thanks. I made a small change to the circuit by adding a resistor between the capacitor and ground. This dampens a resonance in the LF response. The value of the resistor is critical though, and dependent on transformer characteristics and materials. Distortion in simulation at full output is about 0.0003%.
You right Dirk you should pursue what you think it the best for you, Sorry, in future, i will keep my personal opinions for my self. 😉
Hi Dirk,
Based on my limited experience I think you will find yourself iterating as I did - the transformer simulations miss all sorts of transformer subtleties and I ended up with quite different resistor values than I initially expected in order to get transformer damping right, set level, and to keep the core out of saturation - all that said I am quite happy with the end result. (Will share in a new thread when I am ready)
One thought might be to consider combining this transformer concept with a simple discrete current conveyor rather than the op-amp at some point. http://www.diyaudio.com/forums/digi...-sen-evolution-minimalistic-iv-converter.html I have some of these and will be experimenting with them for another dac project.
I'm very interested in how the current implementation turns out.
Based on my limited experience I think you will find yourself iterating as I did - the transformer simulations miss all sorts of transformer subtleties and I ended up with quite different resistor values than I initially expected in order to get transformer damping right, set level, and to keep the core out of saturation - all that said I am quite happy with the end result. (Will share in a new thread when I am ready)
One thought might be to consider combining this transformer concept with a simple discrete current conveyor rather than the op-amp at some point. http://www.diyaudio.com/forums/digi...-sen-evolution-minimalistic-iv-converter.html I have some of these and will be experimenting with them for another dac project.
I'm very interested in how the current implementation turns out.
As long as opinions and suggestions are respectfully expressed you should not hesitate to do so, you might bring something to light, and the dialog might also bring you new insights. You never know - something you suggest might be just the thing another member was looking for. Post away! 😀
Your thread for example has answered a number of questions I had about the PCM1794A saving me a lot of investigation since I do use this part somewhat unconventionally and in a manner where this information becomes highly relevant to me. 😀
I'm limited by lack of information from Lundahl, among other reasons. There are plenty of parameters in the spice model use, but without the data from the manufacturer, I'm left guessing on them.
There won't be any saturation because there isn't any DC current flowing in the windings, at least as far as I can tell.
I consider the transformer is already a "current conveyor" since it feeds into a virtual ground. The I/V resistor could be the feedback resistor of the opamp, since the full AC current output from the DAC flows through it. As I said in an earlier post, I'm more comfortable with IC's than with discrete components.
The way I look at this, the I/V converter is an interface between the DAC and the analog circuitry. So, the only way to test it is with the particular DAC I intend to use with it, and the converter should address the weirdness of the DAC as well (like the -6.2mA of DC current bias on the Iout pins).
Thanks for the suggestions.
Hi Dirk,
Same problem with the Sowter 9545, the specifications were rather vague and recommended implementation was not compatible with the PCM1794A, but I knew this going in, and oddly enough my original thoughts and simulations were closest to where I ultimately ended up, although significantly different - along the way I took a number of dead end detours and learned a lot about what wouldn't work..😀
When I mentioned saturation of the core I was not talking about the DC component in the signal since the 9545 can tolerate 0.6mA of mismatch in the primary windings and my mismatch is substantially less than 100uA.. What I was talking about was the AC behavior of the core, trying to find the best compromise between SNR and overdriving the core such that distortion becomes excessive, at one point I did succeed in driving the cores into saturation which was not very pleasant sounding.. 😛
I'm op-amp adverse which is why I commented, but the op-amp will be a perfectly viable solution in this application, a matter of personal taste and experience.
These transformers are quite a bit smaller than the ones I am using so I'll be very interesting to hear about your results, something similar may be in my future for a small dac for a tiny media server that I am planning around my soon to arrive Raspberry Pi board.
Same problem with the Sowter 9545, the specifications were rather vague and recommended implementation was not compatible with the PCM1794A, but I knew this going in, and oddly enough my original thoughts and simulations were closest to where I ultimately ended up, although significantly different - along the way I took a number of dead end detours and learned a lot about what wouldn't work..😀
When I mentioned saturation of the core I was not talking about the DC component in the signal since the 9545 can tolerate 0.6mA of mismatch in the primary windings and my mismatch is substantially less than 100uA.. What I was talking about was the AC behavior of the core, trying to find the best compromise between SNR and overdriving the core such that distortion becomes excessive, at one point I did succeed in driving the cores into saturation which was not very pleasant sounding.. 😛
I'm op-amp adverse which is why I commented, but the op-amp will be a perfectly viable solution in this application, a matter of personal taste and experience.
These transformers are quite a bit smaller than the ones I am using so I'll be very interesting to hear about your results, something similar may be in my future for a small dac for a tiny media server that I am planning around my soon to arrive Raspberry Pi board.
Well, the LL1517 is an output transformer, so the maximum level before saturation is listed as +24dBu @ 30Hz. I think typical DAC output levels are more like +4dBu. If it doesn't work, well, I'll have learned something.
Current output dacs are bounded by the allowable compliance voltage at the output before the dac linearity is degraded. In the 1794 I take the practical upper limit to be about 100mVpk for this reason. I'm running at about 1/3 of this (including half scale offset..)