Best performance would be gained with jFET input and bipolar cascode. On paper at least.
Thx-RNMarsh
Thx-RNMarsh
Not in my experience. The cascoding device needs to have a higher IDSS and a Vpinchoff high enough to allow sufficient Vds for the Jfet amplifier to work. The MOSFETS work great even with really high Idss parts, which are what you want for low noise.
@1audio,
Which question/comment were you referring to when you answered "Not in my experience"?
Neutrality's or RNMarsch's?
Just to be completely sure.
Which question/comment were you referring to when you answered "Not in my experience"?
Neutrality's or RNMarsch's?
Just to be completely sure.
Actually is was Neutrality's but it applies to both. If you are cascoding to get a low Cin you want the drain to follow the source closely. Biploar devices can do it but you need a follower for the source and then some diodes and a biasing scheme. The benifit of bipolar is the low emitter impedance but the transconductance of the Supertex devices is high resulting in a low source impedance and the threshold voltage gives around 3V gate to source making for an almost ideal implementation. The real challenge is preventing oscillation since the FETs are very fast.
Hi Demian... it doesnt apply equally to both... the bjt can be biased to allow any level you want for the cascode. yes, it is more parts but that isn't in my criteria for this to get best performance from a cascode. Small benefit is that a good low noise bjt can be used without a lot of matching going on. How about those Supertex regarding matching/selecting to get what you want from them? Any serious issues?
The Supertex are super good devices for this and audio. Glad they are still making them.... unlike so many other discrete transistors/parts that have been discontinued.
Just as a point of note: Taking a cue from the Japanese equipment designers, they also use a lot of jFET with bjt cascode.
Thx-RNMarsh
The Supertex are super good devices for this and audio. Glad they are still making them.... unlike so many other discrete transistors/parts that have been discontinued.
Just as a point of note: Taking a cue from the Japanese equipment designers, they also use a lot of jFET with bjt cascode.
Thx-RNMarsh
Last edited:
What is the 'trick' of driving the junction of the diodes at the front end of the SR circuit??
_-_-
_-_-
1) when you don't have the depletion mode fets to try bipolars are the viable option. The depletion NMOS fets are very recent as these things go (10 years or so). They are mostly used in off line mini power supplies. I have never seen an issue with either matching or noise. The on current of the NMOS is way higher than any JFET so thats not an issue. As long as you have gain in the Jfet to the output the noise contribution of the mosfet will be negligible if there. its gain to the output is less than one.
2) the trick is part of the protection circuit. The combination of the diode connected transistors and the light bulbs is the input protection. There are no other clamp diodes at the input etc. The neat part is that the diodes track the input and don't bring non-linearities with them. The lamps are very nonlinear but no current except with stressed. The actual protection is the two lamps, the diode connected transistors and the reverse biased diodes from the sources of the JFETS to the rails.
2) the trick is part of the protection circuit. The combination of the diode connected transistors and the light bulbs is the input protection. There are no other clamp diodes at the input etc. The neat part is that the diodes track the input and don't bring non-linearities with them. The lamps are very nonlinear but no current except with stressed. The actual protection is the two lamps, the diode connected transistors and the reverse biased diodes from the sources of the JFETS to the rails.
Last edited:
The front end is part of the amplifier in the NuForce P9 and P20. I just checked the P9 on my bench (other reasons) and I'm getting all the harmonics at -130 dB or better with unselected LSK389 and DN2535's. It just works.
I will do a test board for the composite with the opamp. That may be harder to make stable. I'm interviewing a new tech tomorrow. That should open up a world of possibilities.
I will do a test board for the composite with the opamp. That may be harder to make stable. I'm interviewing a new tech tomorrow. That should open up a world of possibilities.
Its been awhile and this is an important project IMO. I hope it is still moving along (?)
Whats the latest, guys?
Thx-Richard
Whats the latest, guys?
Thx-Richard
Just a suggestion as seen in the Ardatech AN-AT1201-1 application note figure 9 using an active reference v. cap. based.
Hp
Sounds like a good idea. For the AK5394A we would need two voltages and therefore two op-amps, but that is of course not a problem.
Perhaps it would also provide an opportunity to control the reference voltages more accurately, to ensure a precise and stable reference.
Whether it would create a conflict with the internal bias circuits is something that needs to be confirmed.
Perhaps it would also provide an opportunity to control the reference voltages more accurately, to ensure a precise and stable reference.
Whether it would create a conflict with the internal bias circuits is something that needs to be confirmed.
I also use an active driver with a low noise opamp for the reference.
The input resistance of the circuit with OPA1632 is very low and also has an offset (reference voltage). Because all of my designs are DC coupled, I use a fetbuffer with a dc servo loop. I like the OPA1632 and the sonic performance is great.
The input resistance of the circuit with OPA1632 is very low and also has an offset (reference voltage). Because all of my designs are DC coupled, I use a fetbuffer with a dc servo loop. I like the OPA1632 and the sonic performance is great.
The ESS 32-bit Hyperstream DAC sounds good and i thought, this will be also the best for AD conversion. ES9112 has a DSP and i can load my own filter.
SPDIF/AES output, full configurable for pro and consumer devices out.
External CLK-synchronisation with good pll to reduce the jitter of the incoming clk , very important for recording devices
And i like the sound of the ess stuff, the sound remains analogue and i've never heard that with any other ad-converter.
Best to say, there is no sound.
SPDIF/AES output, full configurable for pro and consumer devices out.
External CLK-synchronisation with good pll to reduce the jitter of the incoming clk , very important for recording devices
And i like the sound of the ess stuff, the sound remains analogue and i've never heard that with any other ad-converter.
Best to say, there is no sound.
Sounds like what I am looking for in an ADC for test and measurement --- 32 bits! and built in DSP.
Thx-RNMarsh
It's 32 bit decimation not sample rate. Sigma delta generates a number of pulses depending on the input voltage level. The pulses are counted and converted or mapped to pcm. This count can be manipulated to output any pcm resolution. They call it a decimation filter.
Or you can just low pass filter the bit stream to convert it to analog. It not necessarily a better way of conversion but it is a whole lot cheaper. SD can't be processed as it is, it has to be converted to PCM.
Or you can just low pass filter the bit stream to convert it to analog. It not necessarily a better way of conversion but it is a whole lot cheaper. SD can't be processed as it is, it has to be converted to PCM.
- Status
- Not open for further replies.