It's easy to measure distortion caused by I/V input impedance using a passive resistor I/V. You can google for people who have done so and published results. I've done some testing myself and have seen a big difference between a 50R resistor and a 2R reflected load from a tube amp output transformer used in reverse.
In the 2SK369V SEN, the Zin is below 10R. I consider that low enough.
If you really believe in super low Zin, then opamp IV is your solution.
Or use (multiple) BJT with a high bias, something like 20mA.
You can see interesting discussions about BJT with Calvin at the start of the thread.
Patrick
If you really believe in super low Zin, then opamp IV is your solution.
Or use (multiple) BJT with a high bias, something like 20mA.
You can see interesting discussions about BJT with Calvin at the start of the thread.
Patrick
Sub-10 ohm is pretty acceptably low, and low input Z is of course only part of the story anyway.
When I read the LA article and was led to this thread, which I've been following ever since, I started looking at some alternatives, that incorporated the SEN approach of floating supplies (which I like a lot) but applied feedback to the SEN input devices' gates from an all-JFET separate gain stage (no opamps per se, no feedback resistors). Other distractions have prevented this from getting beyond those simulations, but the sims looked very promising. Someday...
I went with 12 JFETs in parallel
but the 3rd harmonic was a bit too high so I'm backing off with my 4th Sen instance to 6.(After finally managing to match some JFETs to 0.1%, my 3rd Sen was a tragic fail due to some unset epoxy hardener under the heatsinks - it conducts).
I said most DAC chips, and we are talking about much lower input impedance than 50 ohm with the SEN.
What you say is true. I think it's also true that SEN itself has very low distortion so for some DACs the effect of SEN's input impedance will be the limiting factor.
Hello Patrick. I have a question regarding Sen V18 components values (for es9018 dac).
Can I use for C1 and C2 values bigger or smaller then 220uf and with voltage rating of 10v or 20-25v? (for example can I use something like 470uf 16v or 330uf 10v) And what type is best for C1 and C2, can be used Aluminum Organic Polymer Capacitors for this purpose?
And about Civ. WHat formula should I use to calculate the Civ value. Can be used this calculator at the bottom of page with sign First order RC-filter? And what value should be Fc? I see in original Sen Fc is about 170kHz isn't this bad thing that can bring a lot of RF trash to music? Maybe 100kHz should be fine here?
Can I use for C1 and C2 values bigger or smaller then 220uf and with voltage rating of 10v or 20-25v? (for example can I use something like 470uf 16v or 330uf 10v) And what type is best for C1 and C2, can be used Aluminum Organic Polymer Capacitors for this purpose?
And about Civ. WHat formula should I use to calculate the Civ value. Can be used this calculator at the bottom of page with sign First order RC-filter? And what value should be Fc? I see in original Sen Fc is about 170kHz isn't this bad thing that can bring a lot of RF trash to music? Maybe 100kHz should be fine here?
C1 and C2 I would use minimum 100uF and minimum 16V.
In theory with 18V supply you can use 10V, but I don't like using caps to their rated voltage.
It's just me.
I use non-polar Elna caps for those positions.
Riv and Civ forms a first order RC filter. So just the usual formula.
Whether you use 100kHz or 170kHz or else is a matter of taste.
It also depends on the sampling rate.
I suggest you try different values and listen.
Patrick
In theory with 18V supply you can use 10V, but I don't like using caps to their rated voltage.
It's just me.
I use non-polar Elna caps for those positions.
Riv and Civ forms a first order RC filter. So just the usual formula.
Whether you use 100kHz or 170kHz or else is a matter of taste.
It also depends on the sampling rate.
I suggest you try different values and listen.
Patrick
Thanks, a great article.
In practice I have always used local feedback in BJT based amplifiers. And the article scientifically confirms what's be known for decades: in BJT based amplifiers local negative feedback is quite useful.
Maybe I overdesign, but I like Sziklai pair. With resistor between base and emitter of the second transistor allowing to increase/control first transistor current.
My idea is that it's "infinite" BETA transistor, so using emitter gain degeneration resistor makes very deep local feedback.
For myself I sometimes call Sziklai pair "two transistor operational amplifier"
.I admit I have never simulated it, so I do not know what THD is. On one hand, local negative feedback is very strong, on the other hand, non-linearities are multiplied.
Thanks, a great article.
In practice I have always used local feedback in BJT based amplifiers. And the article scientifically confirms what's be known for decades: in BJT based amplifiers local negative feedback is quite useful.
Maybe I overdesign, but I like Sziklai pair. With resistor between base and emitter of the second transistor allowing to increase/control first transistor current.
My idea is that it's "infinite" BETA transistor, so using emitter gain degeneration resistor makes very deep local feedback.
For myself I sometimes call Sziklai pair "two transistor operational amplifier"
I admit I have never simulated it, so I do not know what THD is. On one hand, local negative feedback is very strong, on the other hand, non-linearities are multiplied.
I like that paper a lot, but be aware of the idealized models that were used. For example, the JFETs have no parasitic resistances in source and drain. When these are included it is as if external R has been added, hence local feedback, hence production of some higher-order distortions.
Sziklai is also of interest. The issues include large-signal asymmetry and significant temperature variations from ambient and signal-induced changes.
... setting up the automated charging circuit is a PIA
I would use relays to connect/disconnect the accumulators to the charging circuit. I mean when the accumulators are being charged the DAC doesn't operate.
E.g. the OFF button might actually mean Recharge button
.Well, yes and no. Nobody in his/her sane mind argues with Kirchoff's Law.
However, we live in EMI jungle. The floating power supply is an RF antenna, and through parasitic capacitances the RF it receives gets fed into the circuit.
You probably know better than I that an ungrounded shield is practically always worse than no shield at all. Your floating power supply is like an ungrounded shield.
Maybe in the final overall shielded I -> V converter it is not a real issue. But one should not blindly adopt floating supplies - everything has its price.
...
If one looks at/into isolated power converters dataheets, he/she can see (e.g. VB-053R3S2 datasheet, Pinout ,application circuits Power Modules DC-DC Converters 2W 2:1 Regulated Single & Dual Output ) : "I/O Isolation Capacitance 60 pF".
I am not sure what you mean. If I go Sziklai, I go Sziklai "all the way". I.e. instead of NPN I use NPN -> PNP Sziklai, and instead of PNP I use PNP -> NPN Sziklai.
Also, as I wrote, I add resistor to increase first transistor current. The resistor is typically 3 .. 6K which allows first transistor to have about 100 .. 200uA (0.64V/R) collector current. I choose these currents as "ballpark" good currents WRT noise.
Decreasing the resistor OTOH increases first transistor collector current improving HF response - not sure I need it though
.I also sometimes use DC servo in my circuits, so DC drifts are taken care of.
...
Because of high effective BETA I also stop worrying about input impedance. For example, with BETA = 200 * 200 = 4e4, and with "effective emitter" gain degeneration resistor Re = 100 Ohm I have DC input impedance of BETA * Re =4e6, which is much greater than any realistic preceding collector resistor. And putting 1K resistor as "effective collector" load I'm getting 10 units of voltage gain with pretty good HF response because 1K is not that big resistance. 1K * 1000pF is still just 1us tau.