| skaara |
I use single ended inpuit (i grounded the other input, but it worked even without grounding it..) for bosoz and convert it to balanced on output, but when i apply sine wave at input i get 1v + phase and 0.7V - phase, i cant figure why (i used matched irf610s, metalfilm resistors...)..
Can anyone help me?:) |
|
|
| jwb |
Sounds like it must be a mistake in the construction. What is the amplitude of the wave applied at the input? It must be awefully small to get a 1Vpp wave at the output, since the BoSoZ without attenuation has 20dB voltage gain.
I would start by checking all of the connections. It is a pretty simple circuit, so first I would make sure that the gate, drain, and source on each FET is wired up properly. Then I would measure all of the voltages with both inputs grounded. Should be 4V below ground at each source pin when idle, if you used +/-60V power rails. Drains should both be about 30V above ground. The idle voltage across 124Ω R15 should be around zero.
If all of that checks out fine, I would remove the output capacitors and measure the peak-to-peak amplitude at each output.
If you still haven't found the problem I would investigate the FETs. Perhaps one of them was subject to ESD. |
|
|
| Nelson Pass |
When converting unbalanced to balanced, there is some
loss on the "undriven" side, which can by corrected by
having greater resistance biasing the diff pair, and gets
practically perfect when you bias the diff pair with a
constant current source. Your other alternative is to
reduce the Source - Source resistance of the diff pair,
which increases the gain also. |
|
|
| skaara |
I used +-60v rails, there is -4v on source of each fet and +30v on drain of each fet, idle voltage on r15 is zero.. What do you mean with removing output cap, removing it and putting normal wire instead of it?
ESD = electrostatic discharge? Can fet be only partially damaged? |
|
|
| skaara |
Got your post Nelson, so I have to use constant current source.. Can you or someone else post schematic of modofied bosoz with current sources then. or just schematic of current source..
I use 1V input which falls if i connect it directly to the amp and i like balanced out too, connected to boz it falls to 0.6v, so there is almost no need for gain.
But if i set the value of r15 to 124ohms i get 3.3V for +phase and 2.6V for - phase, I have a feeling that it should be much more (in the Nelsons article I read that the o/p voltage swing can be 120v at +-60v rails, is that meant for 1v input?)..
And what will be the o/p voltage at, say 20db, of gain for 1V input voltage, how it is calculated? |
|
|
| jwb |
Danger! I mean only to measure the voltage with the output cap removed, to check for a defect there. DO NOT use it without the output caps to drive an amp. Everything will be kaput.
Anyway, it sounds like yours is working in light of the advice from Nelson. My incarnation of BoSoZ uses a contant current source, so I never noticed this assymetry. Apparently it is normal, which makes some sense now that it has been pointed out. |
|
|
| skaara |
| Can you post a schematic of current source? Which resistors does it replace? See my previous post for more questions if you have will:) |
|
|
| skaara |
| Can someone post a schematic of current source and tells me where do I have to put it and tells what are the specs of transistor or fet used because maybe it cant be bought here (i cant get ztxes..) |
|
|
| jwb |
| If you can't get Zetex, use any old BJT. It should be capable of Ic > 50mA, Vce > your supply rail, high Hfe, low cutoff current, and low saturation voltage. You can see the specs for the ZTX869 and ZTX851 transistors at www.zetex.com (or www.zetex.co.uk). I chose them for their small package and ability to dissipate heat. |
|
|
| Henrik |
Skaara
As Mr. Pass said: "When converting unbalanced to balanced, there is some loss on the "undriven" side".
You have a loss at the output at around 30%, it seems too much to me.
I am using my BOSOZ as an unbal. to bal. converter, I do have some loss at "the undriven side" but not more than around 10%, and until now i have not considred this lack of mismatch to be a problem, but i could be wrong. I would like to other oppinions on this question!
The question is, do we get better performance by adding a Constan Current Source, or will the small mismatch from the passive circuit be les significant than an eventually degrade in sonic performance caused by the change from passive to active currentsource.
Grey has "bin there" in a long thread, but that was in relation to SOZ, and since BOSOZ an SOZ do have the same topology, one could get the idear, that his experince could be usefull here.
I my self like the passive topology very much.
You should read Nelsons artickel "DIY OP Amps", you can get from WWW.diyaudio.com .
I have simulated a simple Constant Current Source.
The upper left diagram shows the original BOSOZ, and the one under BOSOZ with CCS, simulated from the circuit at the right. |
|
|
| Henrik |
But here it is WWW.passdiy.com :bawling: :bawling:
I am going to bed now! |
|
|
| skaara |
| But can someone post a sch. of a constant current source which is suitable for bosoz? And which resistor(s) does it replace? |
|
|
| Nelson Pass |
Henrik's current source should work just great.
Thanks. |
|
|
| skaara |
| Which resistors should i replce with current source? |
|
|
| Henrik |
From the org. circiut you should remove P5, R15, R3, R4, R5 and R6.
Replace them with R3, R4, R15, R6, R5, D3 and Q7 from my circiut.
R15, R6, R5, D3 and Q7 is the CCS, R3 and R4 controls the gain. |
|
|
| skaara |
| Thank you, ill do it:) |
|
|
| Henrik |
Skaara
Regarding your first post.
I have done some sims on this.
If R15=124E the loss at undriven vill be around 8%.
If R15=375E the loss at undriven vill be around 20%.
The loss at the undriven increases with lower value for R15.
So if you have chosen the lowest gain and distortion with R15 >= 400E tis could be an explanation or a part of it. |
|
|
| skaara |
| Yes I chosed lovest (10db) gain.. I have some problems choosing the right bipolar transistor, I dont know which of them are suitabe for this (which BCs, BDs or 2Ns?) |
|
|
| Henrik |
In my earlier post I used mosfet IRF610.
You could go bipolar and use TIP31C, very easy to get, but then you have to recalculate the resistor at the emitter to obtain the right amount of current. |
|
|
| skaara |
| What is the formula? Can you please post only schematic of constant current sourcem that I will see exactly what do i have to build:) Thank you:angel: |
|
|
| Henrik |
Skaara
I must be growing wings!
You can use TIP31C or similar. |
|
|
| skaara |
| Sure you re growing wings:) Can I use a diode like 1N4007 (i have a lot of them) instead of MA1068... |
|
|
| Henrik |
Skaara
The diode is a Zenerdiode, and mut be.
Any Zener will do, as long as its voltage is 6.8V, the MA1068 is what i had on hand in my simulater. |
|
|
| skaara |
| 6.8 zener then... These resistor values are calculated for tip31c? |
|
|
| Henrik |
For any Sillicon Bipolar rated at 100V, 3A, 25-50W.
I would prefer a TO220 packet. |
|
|
| macka |
Hi Hendrik & everyone,
I was inspired to try the active ccs gain on my Hendrik X Bosoz this morning.
I recall going back to the resister current source last time after attempting the simple 610/zener circuit.
So I look at a few options in Alain Dupont's design of a 317 negative regulator for 10-15 volt negative supply with a 20uf capacitor across the zenor.
Compared to the resister current sournce the tonal balance was altered a little, more pronounced mids, somehow less very fine detail..less ambience ...a flatter sound.
I then read another post on using a LM329 precision voltage reference and I just happened to have a pair from a super reg project I have not started.
Adusting the 317 to 15 volts, I inserted the LM 329, took out the capacitor and used 1k for about 8 ma through the Lm329 with a source resistance of 34 ohms (68/68R).
My measurements confirmed there is exactly 2.7 V across the source resister to the 15 volt..79 ma.
Well, the sound is wonderful and I cannot hear it apart from the resister ccs. I have no scientific answers other than the LM329 is from the specs, much quieter and more stable with tempco.
My feeling is that the LM329 allows a minimalist approach..less is more.
I might try cascoding the output of the BOSOZ now.....
macka |
|
|
| wuffwaff |
Hi Macka,
so if I understand your report correctly I don´t have to bother testing a current source in my X-BOSOZ since it sounds the same as the resistor version?
I will have a look at the X-BOSOZ in my DAC though cause I believe I´ve used a simple 9V1 zener in here (and no cap paralell)
thanks,
William |
|
|
| macka |
Hi William,
This needs to be verifiied by the Master but as I recall it depends on if you are using the pre amp as an unbalanced to balanced converter. In the is mode the outputs as such are more likely to be equal where an active current source is used.. Previously an active current source altered the sonic signature.
I think in this mode the ccs improves back ground noise and has lower distortion, but best of all the gains in CMRR are not at the loss of sonic qualities.
macka |
|
|
| wuffwaff |
Hi,
I´m using it all balanced. Feeding the X-BOSOZ with ONO and the x-bosoz output stage of my dac (fed by the balanced output of the Crystal chip).
At the moment the x-bosoz in my dac doesn´t even have a regulated supply but easily outperforms the OPA627´s in the regular output stage. I think this will be the next project after finishing the restauration of my Aleph5 and getting rid of the humming transformers in my Aleph-X´s................
William |
|
|
| harryeng |
Hi William,
We are going off-topic, but talking about DAC's:
I also have a X-BOSOZ in my DAC (Kristian/Hendrik). Mine is an AKM DAC which has a balanced voltage output, based on a M-Audio Superdac.
What I am interested in is how you set-up te filtering for your DAC with a Crystal DAC.
I myself upsampled first to 96K, and used a 1-pole filter 3DB at 18Khz plus voltage divider between the DAC and the X-BOSOZ. (three resistors and a capacitor per signal side). According to the specs at 96K input, the AKM needs a brickwall filter at 56Khz.
Greetings, Harry. |
|
|
| wuffwaff |
Hi Harry,
I do use an output filter but with a very high -3dB point. It is based on the original filter used in the P3-A and goes something like this:
1k then 2n2//1k to ground, then 510R and 1n5 to ground, input impedance of the output stage is 27k. This is used for each phase. I remember simulating it but not the exact value. Could be near 47kHz.
William |
|
|
|
