Replacement for 10M45 / CCS

Status
Not open for further replies.
Jump to Latest
#3
hmm, same problem. hard to source for me here in norway. please, could anybody boint me to some reading about this kind of CCS?

seems that a lot of power MOSFET-types could work, but the datasheets are full of slightly divergent characteristics, so i don`t know what to look after.

thanks for any input, harry
 
#5
#8
jcx said:
normal enhancement mode MOSFET require added source of voltage for gate

Welcome to Gary Pimm's DIY Page shows a couple of ccs options

depletion mode MOSFET can work like JFET ccs with source R setting current but there are few part # made
Click to expand...

Hi,

Thanks for your response. BTW - the DN2540 is a depletion mode mosfet. 😱

However, has anyone actually looked at this schematic? There are no external connection points for gate drive. It doesn't appear that anything is required.
So I'm back to the question, will this circuit work, if not, what's needed to make it work? Has anyone successfully used this in a long tail pair? Has anyone successfully modeled this in LTSpice?

I'd be happy to plop this question in another forum if this one's not OK for this, and a moderator could tell me which one.

Thanks,
Gary
 
#10
N type (only type available for MOSFET) depletion mode devcies conduct Idss with Vgs=0, are progressively cutoff by Vgate lower than Vs

the Vgate "comes from" the current flowing in the R in series with the source - allows selecting an operating current that is < Idss

as pointed out this is the same as N jfet ccs, CRD

jcx said:
http://liveweb.archive.org/http://www.vishay.com/docs/70596/70596.pdf

those intersted will probably want all of the Siliconix app notes: JFETs Application notes
Click to expand...
 
#12
jcx said:
N type (only type available for MOSFET) depletion mode devcies conduct Idss with Vgs=0, are progressively cutoff by Vgate lower than Vs

the Vgate "comes from" the current flowing in the R in series with the source - allows selecting an operating current that is < Idss

as pointed out this is the same as N jfet ccs, CRD
Click to expand...


Thanks for both replies. I'll read those Siliconix notes so I have a better understanding of the mechanisms involved. Some folks use just one Dn2540, but the article points out better performance using the cascoded arrangement, so that's what I will try.

In any case, after I last posted, I closed LTSpice and restarted; it now seems to be OK! 😉

Just a couple of further questions.
1 - What should the gate stoppers optimally be for a CCS of about 10mA?

2 - If either of you has modeled this, why does the current through the devices look so funky?

I would use a fixed resistor plus a pot for current trimming.

Thanks again,
Gary
 

Attachments

  • CCS_Current.JPG
    CCS_Current.JPG
    32 KB · Views: 626
#13
The gate stoppers won't play any role in current setting, the source leg resistor does that. You put enough value for gate stoppers until the thing is stable for oscillations in a given app and layout but you don't overdo it because too much makes it slower losing higher band PSRR.
 
#15
jcx said:
cascoding with the exact same device may not give very good results - you really want ~2x Vp across the D-S for the "lower" device to be a decent current regulator
Click to expand...

Thanks again for the replies. I appreciate the sound advice in this area to get me going! 😀
When you say "exact same device", do you mean the same family, or the same lot?

For example, if I were to use a DN2540 in the top of the pole, 10M45S in the bottom, would that be OK since the 10M45S is a CCS device by design?

My cathode AC voltage is 2.65Vp (sitting on 92VDC). So I should seek a device that is linear Idss at that Vds?

Sorry about the questions...learning is asking and the sooner I get the bigger picture, I'll read more and pursue the details to get a better understanding.

Gary
 
Last edited:
#16
roger57 said:
Thanks again for the replies. I appreciate the sound advice in this area to get me going! 😀
When you say "exact same device", do you mean the same family, or the same lot?

For example, if I were to use a DN2540 in the top of the pole, 10M45S in the bottom, would that be OK since the 10M45S is a CCS device by design?

My cathode AC voltage is 2.65Vp (sitting on 92VDC). So I should seek a device that is linear Idss at that Vds?

Sorry about the questions...learning is asking and the sooner I get the bigger picture, I'll read more and pursue the details to get a better understanding.

Gary
Click to expand...

too many questions I guess? sorry to bother, and thanks for the help so far. :boggled:
 
#17
your circuit, diff pair tail ccs doesn't require a self biasing 2 terminal circuit

so unless the "eye candy" factor trumps price/performance, or you simply can't stand still for a "wasted" few 100 uA to bias Vref or regulation circuits you should use one of the many "3-terminal" ccs circuits

likewise BJT have good performance in this app - if you aren't imposing artificial conditions like the silly "only fets sound good for audio"
 
#18
jcx said:
your circuit, diff pair tail ccs doesn't require a self biasing 2 terminal circuit

so unless the "eye candy" factor trumps price/performance, or you simply can't stand still for a "wasted" few 100 uA to bias Vref or regulation circuits you should use one of the many "3-terminal" ccs circuits

likewise BJT have good performance in this app - if you aren't imposing artificial conditions like the silly "only fets sound good for audio"
Click to expand...


I took the time to read the AN103 Siliconix article on CCS devices, and now have a much better understanding.

I have installed right now a power supply fed bipolar CCS, using an MJE340, BC550, and 2 LED reference. It sounds...OK, not remarkable. More detail, but a little stiffer (maybe harsher) but I didn't have a pot in the Rset leg and I really needed to tweak that to see the effect on distortion. I still want to try:
1) the 10M45, and
2) The cascode pair. I've already made the boards in my shop so they're basically read to test. As per your suggestion, I plan to test a different IXYS depletion mode in the lower half of the CCS, but easy enough to use the 2 DN2540's as well to see if I can measure and/or hear a difference.

Before I complete my amplifier re-work, I will also look at a CCS for the input driver stage in the plate circuit. I understand there could be a significant different there as well.

thank you for taking the time to answer,
Gary
 
#19
the dn2540 can cascode the bjt in the LED Vref+BJT ccs - bjt usually work quire well with even 1 V C-E
then you an use lower V, lower power small signal bjt

may still need gate stopper R - have to look with 100+ MHz scope - but often oscillation changes current draw, make it depend on even hand waving parasitic C

I am highly dubious of anyone's ability to "hear" decent ccs, not oscillating - once the ccs performance gets good enough the sensitivity of circuit properties to further ccs "improvement" are small
I would much more believe fft harmonic, IMD distortion plots, frequency sweeps, Audio DiffMaker comparisons Audio DiffMaker
 
Last edited:
#20
jcx said:
the dn2540 can cascode the bjt in the LED Vref+BJT ccs - bjt usually work quire well with even 1 V C-E
then you an use lower V, lower power small signal bjt

may still need gate stopper R - have to look with 100+ MHz scope - but often oscillation changes current draw, make it depend on even hand waving parasitic C

I am highly dubious of anyone's ability to "hear" decent ccs, not oscillating - once the ccs performance gets good enough the sensitivity of circuit properties to further ccs "improvement" are small
I would much more believe fft harmonic, IMD distortion plots, frequency sweeps, Audio DiffMaker comparisons Audio DiffMaker
Click to expand...

Not sure what you mean by "decent" CCS. I guess that means it could be bipolar or FET, both set up correctly.

Lots of individuals touting FET as a better CCS; many of them indicating sound quality. Just maybe they haven't hear a bipolar CCS set up according to what you are saying. I agree that there should not be a difference, but why are there different topologies (not just with CCS, but in general) with one being readily discernible to some and not others?

Because audio is so subjective to the listener, I never dismiss what others say they hear, even if an analysis proves causal or not.

Thanks
 
Status
Not open for further replies.
Top Bottom