In that position it wouldn't really do any more harm than any other transistor. The current variation has almost no effect on output because it is shared equally by the LTP halves. The current variation of this CCS will be almost entirely the base current modulation of Q18 since it's emitter current will be held very constant. If the VAF is about 150 and Beta is 300, 8mA/300Hfe*(1Vin/150Vearly) = about 178nA per 1 volt. 1V/178nA=5.6Mohm. 178nA/8mA=22PPM current variation.
The cascode further mitigates tempco concerns, so I really see no reason to be concerned about the CCS. It's already been far overbuilt.
As for using the KSC1845 in an MC preamp, I am not sure about that. The KSC1845 is not a low-Rb device. In noise it seems to be comparable to the BC550C, but there are much better transistors you could use in a parallel BJT low source impedance situation. Could you have meant the 2SC1815, which actually has a typical Rb of 50 ohms on the original Toshiba datasheet?
The cascode further mitigates tempco concerns, so I really see no reason to be concerned about the CCS. It's already been far overbuilt.
As for using the KSC1845 in an MC preamp, I am not sure about that. The KSC1845 is not a low-Rb device. In noise it seems to be comparable to the BC550C, but there are much better transistors you could use in a parallel BJT low source impedance situation. Could you have meant the 2SC1815, which actually has a typical Rb of 50 ohms on the original Toshiba datasheet?
You're right, except q2, q4, q6 , q8 , q10, q12, q14, q16 aren't 2sc1845 😉
2SC2545 datasheet(3/8 Pages) HITACHI | Silicon NPN Epitaxial
2SC2545 datasheet(3/8 Pages) HITACHI | Silicon NPN Epitaxial
Last edited:
The 2SC254x look like some very nice transistors, but the noise spec of 0.5nV/rtHz is far better than what you could expect from a KSC/2SC1845. That pretty much makes them completely different devices. EDIT: I guess that was your point, although I can only assume that based on what little you write.
Last edited:
Unfortunately that transistor is the only low noise NPN that can put up a fight with the best PNP. Of course unobtainum too from a well known source...The 2SC254x look like some very nice transistors, but the noise spec of 0.5nV/rtHz is far better than what you could expect from a KSC/2SC1845. That pretty much makes them completely different devices. EDIT: I guess that was your point, although I can only assume that based on what little you write.
It would be interesting to see if ZTX851 and friends can maintain low noise at high impedances - I don't think I've seen anything about its current noise.
It's so comical to see so many quotations from The Art of Electronics ...ZTX 1051a and ZTX1055 have better hfe factor yet at 10ma current with 500 hfe you still have a huge current for a mm cart and their 17...50 pf COB makes them clearly usable only with low impedance sources as mc carts.There's no measurement under 1ma where we keep them usually for high impedance circuits...no guarantee of a clear graph at 1uA as with lm394 though...It would be interesting to see if ZTX851 and friends can maintain low noise at high impedances - I don't think I've seen anything about its current noise.
I wouldn't go as far as "comical" but I do find amusement in the loud protestations and pontifications of those who make no measurements themselves.
I used ztx1051 in capacitor multiplier circuits though...I have too many 2sa1038 , 2sc1845, 2sc1740, 2240, to loose my time with unproven theorems for now.I have about 100 ztx651, 751, 851, 951, 1051, 1055...but bc337/327 work so well pretty much everywhere...m5220 too!I wouldn't go as far as "comical" but I do find amusement in the loud protestations and pontifications of those who make no measurements themselves.
0.01pF ?? You are aware actual stray capacitances are orders of magnitude greater than that in the discrete world? femtofarads are strictly an on-chip phenomenon! The feedback network at that impedance is severely band limited.
60dB isn't worth having for an opamp alas, 100dB and up is needed to be realistic, and for low distortion think 120 to 140dB gain as great to have.
Swap those MOSFET current mirrors for BJT Wilson mirrors and you might get a load more open-loop gain? The source resistor values for the current mirrors look far too small to do anything too. A few mA across 10 ohms is a few 10's of mV, the FETs have a few volts of threshold voltage that's competing against, with large variability.
60dB isn't worth having for an opamp alas, 100dB and up is needed to be realistic, and for low distortion think 120 to 140dB gain as great to have.
Swap those MOSFET current mirrors for BJT Wilson mirrors and you might get a load more open-loop gain? The source resistor values for the current mirrors look far too small to do anything too. A few mA across 10 ohms is a few 10's of mV, the FETs have a few volts of threshold voltage that's competing against, with large variability.
Probably a lot more than .01 pF. I'll try to measure it, I have a few 100 Meg resistors around here. The resistor values would set a noise floor that would really limit the utility of the circuit. Apparently its possible to make a good opamp using all MOS but probably not this simple. Input C and miller C would limit the bandwidth a lot.
Maybe a mix of FETs and bipolar could make the circuit much more useful. The output would be very high Z so loading will have a big impact on its performance. Needs followers to deal with that.
Maybe a mix of FETs and bipolar could make the circuit much more useful. The output would be very high Z so loading will have a big impact on its performance. Needs followers to deal with that.
On chip capacitances are measured in femtofarads and even many kilohms of resistance will still look resistive at GHz frequencies in a VLSI chip. In the big world resistors of 1M might start to become non-resistive at 100's of kHz - 100M resistor? Well that's mainly capacitor or inductor at >10kHz....
i could not get a meaningful reading on my bridge. 100000 KHz reads around 2 pF (2.38 pF) with or without the resistor. If there is motivation I'll try another way, otherwise its a lot of hassle for not much knowledge. Fixturing will dominate any measurement and backing out the fixture means a lot of effort. I would need to fabricate something that would allow guarding out the fixture so only the DUT is measured.
That's because your bridge completely dominates for stray capacitance.
Basically I'm just saying a 100M resistor has no place in any audio circuit(*), let alone a low impedance solid state audio circuit, as its impedance isn't constant across the audio band.
(*) Except for biasing condensor mike MOSFET amps, and in high voltage circuitry for electrostatic transducers.
100M resistors are usually not very stable in value or highly linear, as they cannot be made from metal film technology, though there may be some really good highly expensive specialized components out there.
Basically I'm just saying a 100M resistor has no place in any audio circuit(*), let alone a low impedance solid state audio circuit, as its impedance isn't constant across the audio band.
(*) Except for biasing condensor mike MOSFET amps, and in high voltage circuitry for electrostatic transducers.
100M resistors are usually not very stable in value or highly linear, as they cannot be made from metal film technology, though there may be some really good highly expensive specialized components out there.
The P Channel Mosfets seem to have a wrong symbol, they look like N-Channels, just rotated to fit. Not that it matters for reading the schematic, but doesn't that upset the simulator?
Calm down, guys. I wrote the "SOFTWARE MODEL".
The capacitor of 0.01 pF and feedback resistors 100Meg - 110 kOhm - are just WORKAROUNDS to avoid the caveats of modelling software - MicroCap 9, 10. You should know this if you have ever model something in simulation software.
This software model is just a very close approximation of what should be in practice.
Think positively.
About 60 dB gain - I responend earlier.
The capacitor of 0.01 pF and feedback resistors 100Meg - 110 kOhm - are just WORKAROUNDS to avoid the caveats of modelling software - MicroCap 9, 10. You should know this if you have ever model something in simulation software.
This software model is just a very close approximation of what should be in practice.
Think positively.
About 60 dB gain - I responend earlier.
Last edited:
Yes, it is a mistake in a software library of MicroCap 9, 10. Only one of charting of 3 models for BS250 is correct. I will try to fix it later.The P Channel Mosfets seem to have a wrong symbol, they look like N-Channels, just rotated to fit. Not that it matters for reading the schematic, but doesn't that upset the simulator?
> The output would be very high Z so loading will have a big impact on its performance.
> Needs followers to deal with that.
Good advice.
Practically from my experience, the biggest problem will be DC drift.
So the circuitry has to take that into account.
One Last Attempt at Discrete Opamp in DIP8 Footprint
Patrick
> Needs followers to deal with that.
Good advice.
Practically from my experience, the biggest problem will be DC drift.
So the circuitry has to take that into account.
One Last Attempt at Discrete Opamp in DIP8 Footprint
Patrick
- Home
- Source & Line
- Analog Line Level
- Discrete Opamp Open Design