Hi Patrick,
Can you explain further? (Or did I miss an earlier post?) The four mosfets I used all matched pretty closely, but I used the closest pairs for the two channels, rather than using them in "tops" and "bottoms", which it seems you say is better... I'm wondering if it's worth switching them over. How close would you normally hope to match?
Cheers
Nigel
In the standard configuration, the bottom JFET + MOSFET is just a constant current source. So the output characterisitcs is mainly determined by the upper FETS. Since the cascode LU1014 circuit makes use of signal dependent changes of Vds to achieve the triode effect, the cascode MOSFET becomes an (semi-)active part of the "triode mode" circuit. And since we want equal performance on L&R channels, hence the match L2R.
In the TCS configuration, the bottom FETs form part of the push pull. In that case, both groups in each channel should be matched, i.e. you need a quad set of 1914's and a quad set of MOSFETs.
And if you really want best performance, you should not only match Vgs at working Id, but also transconductance -> curve tracer match.
Patrick
In the TCS configuration, the bottom FETs form part of the push pull. In that case, both groups in each channel should be matched, i.e. you need a quad set of 1914's and a quad set of MOSFETs.
And if you really want best performance, you should not only match Vgs at working Id, but also transconductance -> curve tracer match.
Patrick
do you mean match Vgs at different Id?
what audible difference can this further matching make?
OK, that makes sense to me, thanks. You didn't say how closely you would expect to match the mosfets L2R. In one of Nelson's articles he says to hope for 0.01V variation, and settle for 0.1V if necessary. In the case of the DAO (without Taylor) what would you reckon is "close enough"?
Cheers
Nigel
L-R matching only helps soundtaging and imaging so what is tolerable is up to you. also if your preceding stages are not closely match it doesnt help to have Vgs+/-.00000001%.
that said i d guess within 10% is ok but even lower is better. someone correct me if i m wrong
Last edited:
LR match affects sound stage if it is just an amplitude issue, like in a stereo attenuator (volume control).
It this particular case, it affects linearity (distortion).
I am obsessed with matching, but I also know practical limits.
For mosfets, 10mV is a good indicator.
Any less than that you are into multimeter drfits, MOSFET temperatures, ......, etc.
And you should really match at heatsink temperature.
(We match ours inside an oven at 55°C.)
So if someone tells you he matches to 1mV and does not charge you 10USD extra for the work, he does not know what he is doing.
Patrick
It this particular case, it affects linearity (distortion).
I am obsessed with matching, but I also know practical limits.
For mosfets, 10mV is a good indicator.
Any less than that you are into multimeter drfits, MOSFET temperatures, ......, etc.
And you should really match at heatsink temperature.
(We match ours inside an oven at 55°C.)
So if someone tells you he matches to 1mV and does not charge you 10USD extra for the work, he does not know what he is doing.
Patrick
Hi Patrick,
I have no problem understanding the soundstage/amplitude issues in LR matching, but how can LR matching affect linearity? This has me really puzzled...
Cheers
Nigel
@pidesd
You are mixing things up.
Post #242 & 246 refers to the necessity to match the cascode MOSFET, not the LU1014.
The vgs of the cascode MOSFET will affect the Vds seen by the LU1014 1:1, and hence is important, as the "triode effect" only occurs at a certain Vds. Please refer to Nelson's article on Zen version 9 for explanation.
Of course the Vgs match of the LU1014 is also important, as it changes the bias current, since the source resistor x the bias current provide for the negative Vgs as shown in the figure in post #26. As long as it is not far off from Id =0.2A when Vds=3V & Vgs = -1.35V, graph 26 does not change sigificantly.
See also :
http://www.diyaudio.com/forums/headphones/95841-mosfet-follower-headphone-amplifier.html#post1130939
I measure & fine match my JFETs which I bought ages ago from Grey Rollins pre-matched, so I have no idea or simple means to convert those Deepsurplus values to graph 26. Only thing I can suggest is to buy some with a mid-value, and do some measurements yourself to confirm.
Patrick
You are mixing things up.
Post #242 & 246 refers to the necessity to match the cascode MOSFET, not the LU1014.
The vgs of the cascode MOSFET will affect the Vds seen by the LU1014 1:1, and hence is important, as the "triode effect" only occurs at a certain Vds. Please refer to Nelson's article on Zen version 9 for explanation.
Of course the Vgs match of the LU1014 is also important, as it changes the bias current, since the source resistor x the bias current provide for the negative Vgs as shown in the figure in post #26. As long as it is not far off from Id =0.2A when Vds=3V & Vgs = -1.35V, graph 26 does not change sigificantly.
See also :
http://www.diyaudio.com/forums/headphones/95841-mosfet-follower-headphone-amplifier.html#post1130939
I measure & fine match my JFETs which I bought ages ago from Grey Rollins pre-matched, so I have no idea or simple means to convert those Deepsurplus values to graph 26. Only thing I can suggest is to buy some with a mid-value, and do some measurements yourself to confirm.
Patrick
@njepitt
The graph in post #26 & the link in post #249 is set such that Vds across the LU1014 at DC bias is about 3V. If that value deviates significantly, the graph in post#26 will change. The Vds of the LU1014 is set by the source resistor, the zener diode and the Vgs of the MOSFET together. A different Vgs of the cascode MOSFET will change the Vds seen by the LU1014 (and without saying you should also match the Zener diodes to the same degree). And Vds of the 1014 will change its triode mode characterisitcs.
By how much ? How senstivie ? I have not investigate all permuatations, so I do not know. I just avoid mismatch in the first place. Since I always buy transistors in bulk (25 pcs minimum) and I curve trace every one, I just need to look up my records to find what I want.
I know it is not of much help to you, but then you also might not need to chase the last 0.1% as I do. So my advice is match as well as you practically can, build, and then decide whether it is good enough in the first place to start thinking about chasing perfection.
Patrick
The graph in post #26 & the link in post #249 is set such that Vds across the LU1014 at DC bias is about 3V. If that value deviates significantly, the graph in post#26 will change. The Vds of the LU1014 is set by the source resistor, the zener diode and the Vgs of the MOSFET together. A different Vgs of the cascode MOSFET will change the Vds seen by the LU1014 (and without saying you should also match the Zener diodes to the same degree). And Vds of the 1014 will change its triode mode characterisitcs.
By how much ? How senstivie ? I have not investigate all permuatations, so I do not know. I just avoid mismatch in the first place. Since I always buy transistors in bulk (25 pcs minimum) and I curve trace every one, I just need to look up my records to find what I want.
I know it is not of much help to you, but then you also might not need to chase the last 0.1% as I do. So my advice is match as well as you practically can, build, and then decide whether it is good enough in the first place to start thinking about chasing perfection.
Patrick
Last edited:
On the contrary, it is a very great help indeed, and I'm grateful for the time you have taken to explain things - after all, part of the point of diy is to have fun building things and enjoying them afterwards, but part of the fun is also to try and understand.
In practical terms, while I would certainly "chase the last 0.1%" if I had the resources, I will have to settle for measuring voltages everywhere in the working circuit and see if there is any point in swapping two mosfets over, so the best pairs are left-to-right, instead of top-to-bottom...
Thanks again
Nigel
so if i understand clearly, we want to see a lu1014 Vds of around 3v to have a ''triode effect'' (i will read that zen article). to do so, we want a lu1014 with a certqain Vgs (a middle value Vgs from deep surplus should be ok) that acts in it s linear region at 200ma bias like in post 26 AND we want a mosfet with an unknown Vgs that will ''put'' the LU1014 Vds at 3V also at the same 200ma. so if one of the Vgs value of any transitor in the upper cascode changes just a little bit, then when we want to duplicate the circuit for the other channel, the triode effect might be lost.
is that it?
so how do i know what mosfet Vgs will give me 3V Vds on the lu1014 before i make the circuit?
with the mosfet Vds given by:
Vds=20V - Vlu1014 - Vsource resistor
all this at 200ma?
is that it?
so how do i know what mosfet Vgs will give me 3V Vds on the lu1014 before i make the circuit?
with the mosfet Vds given by:
Vds=20V - Vlu1014 - Vsource resistor
all this at 200ma?
Last edited:
Yes, your understanding is correct.
Of course if you know all values, you can either pick a zener of the right voltage or use a trimmable resistor to replace the zener so as to set the Vds of the 1014 by hand.
The equation is :
Zener (or resistor) voltage = 1014 bias current x 3R + 3V (1014 Vds) + Vgs mosfet at the same bias current.
AND
the 1014 bias current occurs at a Vgs = -(I bias x 3R)
When using a resistor to replace the zener, the voltage across it is of course ohm's law,
i.e. V "zener" = i (CRD) * R
Patrick
Of course if you know all values, you can either pick a zener of the right voltage or use a trimmable resistor to replace the zener so as to set the Vds of the 1014 by hand.
The equation is :
Zener (or resistor) voltage = 1014 bias current x 3R + 3V (1014 Vds) + Vgs mosfet at the same bias current.
AND
the 1014 bias current occurs at a Vgs = -(I bias x 3R)
When using a resistor to replace the zener, the voltage across it is of course ohm's law,
i.e. V "zener" = i (CRD) * R
Patrick
Building something first.
All other guys who built it did not match to the extreme.
All seem more than happy with the results.
When you are so happy with the sound that you want to chase the last 0.1%, you can always contact me for special (-> expensive) Q sets.
Maybe you don't like it at all ..... who knows.
Patrick
All other guys who built it did not match to the extreme.
All seem more than happy with the results.
When you are so happy with the sound that you want to chase the last 0.1%, you can always contact me for special (-> expensive) Q sets.
Maybe you don't like it at all ..... who knows.
Patrick
Maybe you don't like it at all ..... who knows.
Patrick
with all the hype going on and all the effort i m putting in, it better be good! if ever it is not head and shoulders above everything else i have tried, i will throw the damn thing down the toilet.
seriously, it is vey kind to help me understand the circuit. very much appreciated.
Last edited:
Anyone rate AKG K530?
Hi Guys,
I didn't really expect to use the DAO much until on sabbatical leave in January, but I find I'm listening to it a fair bit using the cheapo Maxell phones I bought to check it worked. Although they sound better than I expected, they're obviously not really up to scratch, so yesterday I went looking to see what I could find. I should point out that headphones here are sold mostly in shops selling musical instruments and pro-audio/DJ type equipement, so it isn't practical to listen to the phones before buying. Also, this being Brazil, all the prices are astronomical compared to the US or Europe.
The only phones I found that are worth considering are Sennheiser HD202, AKG K414P and AKG K530. The Sennheisers are apparently $25 in the US, so at $120 here I feel they are too overpriced. The AKG K414P are about $45 in the US, about $110 here, which isn't *too* bad, but the reviews I read suggest they are bass-heavy, and best for rock rather than the classical I listen too. The AKG K530 are a much more serious item; they cost about 76 pounds in the UK, about 175 pounds here, but are very comfortable and seem to have reasonable reviews. Either of the AKGs could serve as second phones, maybe at work, once I buy better phones in the US next year, and would give me better music right now On the other hand, it's a a lot of money to waste if they'll be disappointing, or redundant...
So: the specs say the K414P is 32R impedance, the K530 is 55R impedance, which from the posts above suggest they should match the DAO reasonably well, but am I overlooking anything? (I don't expect the performance to be as good as more expensive cans) Does anyone have experience with them?
Thanks for any info
Cheers
Nigel
Hi Guys,
I didn't really expect to use the DAO much until on sabbatical leave in January, but I find I'm listening to it a fair bit using the cheapo Maxell phones I bought to check it worked. Although they sound better than I expected, they're obviously not really up to scratch, so yesterday I went looking to see what I could find. I should point out that headphones here are sold mostly in shops selling musical instruments and pro-audio/DJ type equipement, so it isn't practical to listen to the phones before buying. Also, this being Brazil, all the prices are astronomical compared to the US or Europe.
The only phones I found that are worth considering are Sennheiser HD202, AKG K414P and AKG K530. The Sennheisers are apparently $25 in the US, so at $120 here I feel they are too overpriced. The AKG K414P are about $45 in the US, about $110 here, which isn't *too* bad, but the reviews I read suggest they are bass-heavy, and best for rock rather than the classical I listen too. The AKG K530 are a much more serious item; they cost about 76 pounds in the UK, about 175 pounds here, but are very comfortable and seem to have reasonable reviews. Either of the AKGs could serve as second phones, maybe at work, once I buy better phones in the US next year, and would give me better music right now
On the other hand, it's a a lot of money to waste if they'll be disappointing, or redundant... So: the specs say the K414P is 32R impedance, the K530 is 55R impedance, which from the posts above suggest they should match the DAO reasonably well, but am I overlooking anything? (I don't expect the performance to be as good as more expensive cans) Does anyone have experience with them?
Thanks for any info
Cheers
Nigel
No, and no, I'm afraid... If I buy online (which I have done with many less bulky objects) I have to pay for the more expensive shipping (usually FEDEX or UPS or similar) to be sure it arrives safely, and then have to pay 70% import taxes on the value of the purchase *and* the shipping... So for headphones like this you can expect to pay close to the prices I listed, and have no warranty at all.
I'd do second-hand happily, but it's only really practical in Sao Paulo or maybe Rio, not here in Brasilia - I'm wary of using the Brazilian version of e**y.\
This is why I'm waiting to buy K702s or something equally "good" when up in the US.
Meanwhile, any input on the likely sound fro the K414s or K530s?
Cheers
Nigel