Zen -> Cen -> Sen, evolution of a minimalistic IV Converter

Assume you would like 2Vrms per phase, as most standard CD players.
2Vrms is approx. 5.6 p-p.
I hope it isn't too much of a distraction to go through this, but 2 Vrms is +6 dBV where 1 Vrms would be 0 dBV. A lot of consumer gear has -10 dBV output, or 0.316 Vrms. Seems like a 16 dB increase (from a consumer DAC to a CD player) is a lot, but I know you're right that CD players output "2V" - I just cannot recall whether they meant 2 V p-p or 2 Vrms.

Are you sure that the average CD player is not outputting 2 V p-p? That would be 0.776 Vrms or -2.2 dBV, which is not quite so extreme beyond the consumer -10 dBV standard.

I understand that car stereo preamps are outputting 4 V these days, but again I have no references on whether they mean 4 V p-p or 4 Vrms.

If anyone can clear this up, I'm all ears. I'm hoping that my dBV/Vrms/V p-p conversions are correct.

In any event, the math given by Patrick will obviously work no matter what you choose as a level standard.
 
> Is my understanding correct that I should use unipolar 18V voltage source and not a bipolar +/- 18?

Yes. The original SEN / CEN circuits have to make use of the floating supply to guarantee low distortion.

> I also wonder what's the point of doubling JFETs in Sen vs Cen.

Because the lower 2 JFETs are constant current source and do not carry signal.
So in order to obtain the same low distortion and also 15R input impedance, you need to double the devices.
You can also use a pair of matched 2SK369 to replace 4x 2SK170s.
But 2SK369s are probably not as easy to get.

You should read the article in Linear Audio.

> Are you sure that the average CD player is not outputting 2 V p-p?

I am sure you can also use google "cd player output standard 2v rms" :

cd player output standard 2v rms - Google-Suche

Marantz Reference SA-KI-Pearl SACD/CD player Measurements | Stereophile.com

".... the maximum output level is a very audible 2dB higher than the CD standard's 2V RMS."

DIN HiFi Vintage Audio Interface Matching: Connections used in the 60s & 70s are Different from Today's Levels | Suite101.com

"Typical levels are

* Line Output level (non CD players) 0.775Vrms
* CD player Output Level 2V rms
* Output impedance < 600 ohms"

> Would be interested in a kit, if you're still pursuing it!

Yes.

While we have ordered further PCBs of Calvin's and Joachim's circuits, simulations showed that the original circuits are still very respectable in distortion levels in comparison, are very quiet, and you cannot beat them on pure simplicity. We still intend to use the SEN IV in our own SD card players.

On top of that, the other circuits belongs to the other two gentlemen.
So you should contact them if you want PCBs for those. And not us.

So we still intend to start a GB thread for the evaluation kit around mid September.
See if we can reach the minimum quota of 60 sets .... ;)


Patrick
 
> Are you sure that the average CD player is not outputting 2 V p-p?

I am sure you can also use google "cd player output standard 2v rms" :

cd player output standard 2v rms - Google-Suche

Marantz Reference SA-KI-Pearl SACD/CD player Measurements | Stereophile.com

".... the maximum output level is a very audible 2dB higher than the CD standard's 2V RMS."

DIN HiFi Vintage Audio Interface Matching: Connections used in the 60s & 70s are Different from Today's Levels | Suite101.com

"Typical levels are

* Line Output level (non CD players) 0.775Vrms
* CD player Output Level 2V rms
* Output impedance < 600 ohms"
Thanks for indulging my questions, Patrick.

I'm old enough to remember the introduction of the CD (which, of course, I boycotted in favor of vinyl until years later when the quality of digital came up to par), but I recall the folks in the stereo shop saying that the CD output was twice the standard line level. Apparently the increase was significantly more than twice the non-CD norm. So many people quoted "2V" without specifying RMS or peak-to-peak. Of course if you search for "rms" then you'll find quotes that claim that to be the standard.

I'm not one of those people who refuses to believe anything on Wikipedia, but I will say that I've occasionally found some serious errors online. There are so many technological myths that we sometimes find that the internet documents the myths as truths. It doesn't help when so many of the participants are too young to remember the previous technology or even the introduction of the current technology. I corrected one very serious error about the CD format (where my memory was more accurate than above, and backed up by printed documentation) where the erroneous value was repeated so often that everyone simply assumed it was correct, having never been around when the information first surfaced for the perspective that brings.

In any case, I trust John Atkinson.
 
Hi,

I really admire the simplicity and the fidelity of those JFET circuits.
Very low noise, very low distortion, and a nice THD-spectrum. One couldn´t ask for more. The reasons to evaluate the bipolar version are cost and sourceability of parts, especially complementary parts.
In the simulation process it showed, that the bipolar version performs very similar to the JFET versions. Distortionwise on par (You can ´shape´ the THD-spectrum with the choice of idle current) and noisewise very close and more than sufficient (<<-110dB). The bipolar version features a considerably lower and more linear Zin. Zin greater than app. 20Ohms increases the THD of DACs like PCM1792/94 et al and the Zin should be linear up to at least the output clock frequency of the DAC (several MHz).
The most noticeable drawback of the bipolar is its need for a biasing network.

Regarding the DAC centre current of BB/TI DACs. The BB/TI DACs current orientation is a bit confusing, because from a current sourcing DAC one might rather expect a positive signed current. Then, EUVL suggestion would apply. But this is not the case. The -3.5mA mentioned in the Datasheet of BB/TI mean that the current is not sinked but sourced, i.e flowing from the DAC to the load. Using a SEN-circuit, the lower, constant current source JFETs need to take the DAC centre current and the upper JFETs idle current. The upper JFETs only takes the signal current.
If You´re using a CEN-circuit, a dedicated current sink is needed, referenced to a negative supply, to null out the DAC´s centre current.
The current sink is best a N-JFET trimmed to the DAC´s current value with a source resistor/pot. Trimming the current sink, You can at the same trim the drain voltages to symmetrical Vdc/2 (differences in the gm of N- and P-JFET lead to a asymmetry of the drain voltages).
Sidenote: The noise of the current sink will be the dominating noise source of the complete circuit. So it needs to be a very lownoise source. Funny, but a simple N-JFET current sink performs much better noisewise than more stable and more complicated CCS. So even the bipolar version might at least feature one JFET. Simplicity wins again ;)

When using DACs with considerable signal current swing (PCM1792/94, especially in Mono Mode) driving into a SEN, You should check if the JFET can handle the increased stress.
Noisewise and for lowest Zin it´d be best to omit with source resistors alltogether, running the JFETs at their Idss. But, depending on the supply voltage this might already lead to a lot of heat through power dissipation. Now the signal current adds to Idss for half the signal period, driving the gate positive against the source. A working condition usually not shown in the Datasheet, where the diagrams end at Vgs=0/Idss). Only small values of positive Vgs may be allowed. Check for this and that the power dissipation does not exceed the limits of the transistors.
Source resistors to reduce the drain current/power dissipation and to distribute currents even among paralleled transistors may be needed (up to 10-15Ohms with acceptable noise penalty and still acceptable low Zin).
The CEN is much less critical here, since both -the upper and lower signal path- take just half of the signal current and a DACs centre current is sinked away by the dedicated CCS.

jauu
Calvin

ps. the bipolar version ceratinly does not belong to me. I was just the guy who caught the bouquet :D
 
Last edited:
Calvin,

Thanks for the info regarding the BB Sigma Delta DACs output current bias.
I don't use any of them, so my knowledge is not more than datasheets.

I should have known that they are current sourcing, as the DC output of the IV opamp is -ve.

In the CEN, one doesn't have to add an N-FET current drain, theoretically speaking.
One could just choose a P-JFET that has 3.5mA less Idss than the N-JFET.
Say for example 10mA for 2SK170, and 6.5mA for 2SJ74.

I know I am quoting the two extreme ends of the BL grade current range.
So in the end the additional FET is the pragmatic solution.

As to coping with high DAC output currents, I recommended increasing the number of devices accordingly in my article.
So the original CEN with one pair will do for 4mA pk-pk i_out, and if you have 8mA, you would need 2 pairs.
The PCB will allow that.

If you want to use the SEN for 8mA, I suggest you use 2 pairs of matched 2SK369V (14~16mA Idss if possible).
They will be OK for dissipation as long as you do not go much further than 18V for the floating supply, and you use a heat sink on top.

The discussions with you and Joachim have been very enjoyable.
It certainly makes the thread a lot more lively.
Your inputs are most appreciated.

;)


Patrick

.
 
Last edited:
Vout DACs are normally already at 2Vrms out. So all you need, if at all, is a basic follower.
Even that is arguable, as the output impedance of opamps are usually very low.
Whatever "damage" you might have done with the internal opamps, you cannot correct for with an additional follwer.
Your only other option is to use a Iout DAC, IMHO.

If you are looking for a follower, I have another thread at the solid state forum on a compact follower based on 2SK372s.
Cannot be more simple.


Patrick
 
> Is SEN your current ultimate sounding I/V or It wins because of simplicity

It is what I shall be using myself.
I don't use words like best or ultimate. Else I can stop.

> does the kit include floating supply or could you recommend such

I use 16x AA sized NiMH batteries with a electrolytic cap in parallel.

> would you sell balanced kits

Each PCB will support one balanced or 2 single ended channels.
So each evaluation pack already contains 2 channels of balanced SEN & 2 of CEN.


Patrick