The Borbely designs usually look good on paper but the component count is a bit over the top for me. I have built and listened to just one of them some 20 years ago and it did not impress me at all except for noise, it was dead quiet. Dull someone said and i could not agree more! It made the otherwise so vivid SPU lose most of its strong points to my ears. I did make some design choises in the circuit in post #399, no parallell devices to keep it mm compatible, ccs load for the input devices as opposed to resistors to make the demands on the ps lower. I do think it would perform quite well in practice for most carts except really low output mc´s. If i would actually build it i would use other led´s for ccs set up and most probably 2sa1316/2sc3329 for all bjt´s.
BR,
Anders
BR,
Anders
Hi Ingemar,
Cascoding helps, but is only part of the story. Yes, it does remove the Miller effect in those topologies where that comes into play. However, even with a cascode the drain-gate capacitance is still there, but now just to AC ground. The drain-gate capacitance of a 2SK170, for example, is 6 pF at Vdg = 10V. It rises to 10pF at 3V and 15pF at 1V. If you have 4 of these in parallel, this nonlinear capacitance can be as high as 40pF in a design that is cascoded but with only 3V Vdg on the device to keep the noise down.
If you have a total of 8 devices, four N and four P in a complementary design, the capacitance becomes 80pF.
Using a driven cascode, where the bases of the cascode signals are driven with a replica of the gate signal can mitigate the effect of the Cdg capacitance.
The nonlinear gate-source capacitance also cannot be ignored in many designs. Cgs for the 2SK170 is about 30pF. Eight such devices in a complementary design makes a whopping 240pF.
Designs that use negative feedback to the input stage will reduce the effective Cgs by tending to move the sources with the signal - a form of bootstrapping.
Cheers,
Bob
Hi tvi,
Thanks for doing this googling and presenting these results. It looks to me like there is a pretty good spread, although it would be more complete if the nominal output levels on most of these were specified. In particular, it would be interesting to know the output voltage for the Kiseki, as it might be a high-output design. Similarly, the Denon DL304 at 50uH and 40 ohms might be a high-output design.
Cheers,
Bob
Go for it, IngemarR. Build it and love it. Better yet, buy the kit, if they are still available.
I build with the SAME parts, even today.
However, we have 50,000 already in house, and an inside track to vendors. Charles Hansen, my esteemed competitor, purchased 500,000 J74 devices for his future use. Perhaps you have access to a few to put something together and would like to share them with your friends and colleagues. '-)
I build with the SAME parts, even today.
However, we have 50,000 already in house, and an inside track to vendors. Charles Hansen, my esteemed competitor, purchased 500,000 J74 devices for his future use. Perhaps you have access to a few to put something together and would like to share them with your friends and colleagues. '-)
Which is even more non-linear with voltage than the capacitance, and it varies with both drain current and Vgd.
Bob - I'm surprised at your statement. Compute the nonlinearity caused by say a 1mV MC signal (with Rs = 10 Ohms) on top of a -1V Vgs with a 240pF capacitance with standard 1/3 power voltage variation. Then compare it to the open-loop distortion of the FET.
Hint: 240pF becomes 240.072pF
I get ~14ppm @ 10kHz
Last edited:
Yes, gate leakage of 10 pA at 10V. Wow!
For the record, does anyone else actually look at a data sheet? They are available still, and they can answer many questions, especially when second guessing other people's designs. Usually the designer has looked at the data sheet and tries to avoid obvious pitfalls.
For the record, does anyone else actually look at a data sheet? They are available still, and they can answer many questions, especially when second guessing other people's designs. Usually the designer has looked at the data sheet and tries to avoid obvious pitfalls.
Yes. The LSK170 datasheet shows 1nA, not 10pA at an unspecified current and Vds = 0. Toshiba sheet for 2SK170 shows similar number. Where did you come up with yours?
Neither describes the actual operating conditions for the FET in this circuit.
The Kiseki was low output, as all Kisekis were. They used another brand name, Milltek, for their (cheaper) high output MCs.
The Denon DL-304 is most certainly low output: 0.18mV.
That graph isn't on their datasheets. What are the test conditions, specifically Vds and id?
It is really simple. They apply a voltage between the source and the drain. Then they measure the gate leakage, apparently at Idss, or they may have another DC voltage to control Id. They call it Igx-Vds.
I just spoke to a manager at Linear Systems, and they claim to meet or BEAT the leakage spec of the 2SK170, and we discussed that something might be wrong on their spec sheet.
I just spoke to a manager at Linear Systems, and they claim to meet or BEAT the leakage spec of the 2SK170, and we discussed that something might be wrong on their spec sheet.
I have the datasheet resulting from a Google search. It specs gate leakage at Vds = 0, but with varying Vgd, which ignores the mechanism that creates excessive gate leakage at high Vds. If you have something which has more info (especially ig at Vds = (say) 10V and id at 1ma, 10ma, or something like that), I'd love to see it. Email addy is my initials at syclotron.com.
What I was trying to say, John, was that it is possible for DIYers to get some 2SJ74:s even today. A DIYer does not need to have 1000s of JFETs. Thus, most JFETbased amps can still be built.
Erno has closed his business last year already.
Erno has closed his business last year already.
Werner,
The "bird" I believe goes to Ortofon, MC7500 is 0.12mV and the MC-1000 & 2000 where 0.1mV and 0.05mV (not certain which way around).
The MC7500 is the cartridge I get to Demo my Phono with, in Hong Kong and China (as the distributor carries only Ortofon and Benz and the MC7500 is pretty much the only one that sounds good on that NA Table they give us).
I find a Phono Stage with -145dBV Ein (e.g. my one) works fine with the MC7500 and the 0.1mV Ortofon, but I have not tried the 0.05mV pickup.
Ciao T
The "bird" I believe goes to Ortofon, MC7500 is 0.12mV and the MC-1000 & 2000 where 0.1mV and 0.05mV (not certain which way around).
The MC7500 is the cartridge I get to Demo my Phono with, in Hong Kong and China (as the distributor carries only Ortofon and Benz and the MC7500 is pretty much the only one that sounds good on that NA Table they give us).
I find a Phono Stage with -145dBV Ein (e.g. my one) works fine with the MC7500 and the 0.1mV Ortofon, but I have not tried the 0.05mV pickup.
Ciao T
Hi Scott,
The main point I have been making is the concern about device capacitances in use with MM cartridges, where the impedance levels are much higher, and indeed the signal levels are much higher. My main point is that input stages designed for extremely low noise by using multiple paralleled JFETs are not a good idea for use with MM cartridges.
Your calculations with Zs = 47k and 5mV signal level will probably yield a bigger number.
Cheers,
Bob
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.