Is there any advantage of using jfets in the LTP than just using bipolar transistors with current mirror with dc offset balance pot and the cussent source/sink ?
Or was it just something they did back in the days before they fully figured out the current mirror and that kinda stuff ?
Or was it just something they did back in the days before they fully figured out the current mirror and that kinda stuff ?
Member
Joined 2009
Paid Member
never tried 'em but I see tradeoffs whether jFET or Bipolar. Things I've considered as different between them and relevant for LTP - breakdown voltage, transconductance, input capacitance, ease of matching, availability, input headroom, temp.co
The number one reasons for using JFETs in the diff pair are high input impedance and low noise.
Depending on how golden your ears are (and/or how deluded you are 😀), there is probably not an audible difference when using low noise/high gain BJTs.
Depending on how golden your ears are (and/or how deluded you are 😀), there is probably not an audible difference when using low noise/high gain BJTs.
Here I must agree , a good low noise BJT like the mighty MPSA18 will sound the same as even a TO-5 matched "audiophool" Jfet pair. I have done this on my "sacrificial" AX modules. NO sonic improvement with the JFETS , In fact , I could not tell the amps apart even after tweaking the tail current and altering the degeneration to give both sets of devices the best operating conditions.
The Jfets had worse offset than the BJT's , but this was due to the differences in degeneration to set the amp to spec. A MAJOR disadvantage is the COST of the Jfets.... 🙁
OS
The Jfets had worse offset than the BJT's , but this was due to the differences in degeneration to set the amp to spec. A MAJOR disadvantage is the COST of the Jfets.... 🙁
OS
Borbely would only use jFETs to the exclusion of all super spec BJTs.
Curl talks a lot about the advantages of using jFETs.
Cordell also mentions some of the jFET advantages.
Two devices costing £0.30 to £1.20 each jFET vs £0.05 to £0.60 each BJT hardly seems to be an issue in an amplifier exceeding £100.
Curl talks a lot about the advantages of using jFETs.
Cordell also mentions some of the jFET advantages.
Two devices costing £0.30 to £1.20 each jFET vs £0.05 to £0.60 each BJT hardly seems to be an issue in an amplifier exceeding £100.
In an LTP application of a CCS, much of the criticality is removed because of the low impedance at the junction of the sources/emitters/cathodes. This minimizes the effects of mediocre source impedance or capacitance since the lower leg of the effective voltage divider is low impedance. Since it's easy to do a really high performance CCS there, I'll do that anyway, but really, there's just not that much difference between something simple and something fancy at that spot.
Now as collector/plate loads, that's a different story... bring out your best.
Now as collector/plate loads, that's a different story... bring out your best.
Well, using FETs makes sense with instrumentation amplifiers (i.e. biopotential amplifiers) where you are dealing with impedances on the hundreds of KOhm and potentials in the mV in the best case.
Here, (Power Amplifiers) with impedances in the order of the hundreds of Ohm and potentials in the order of the Volt, I do not see real avantages. It can be done with good results, of course, but it is far from necessary
and advantages must be traded with disavantages.
Just my 2 Lire.
effebi
Here, (Power Amplifiers) with impedances in the order of the hundreds of Ohm and potentials in the order of the Volt, I do not see real avantages. It can be done with good results, of course, but it is far from necessary
and advantages must be traded with disavantages.
Just my 2 Lire.
effebi
With JFETs in the LTP you generally can get rid of the Input and Feedback caps. This in itself will make a sonic difference, all other things apart; No Cap is better than "Any" Cap at all.
Sorry, but, generally speaking, I do not think that it is safe.
I believe you should always set up an high pass filter (i.e. DC blocking) on the input.
What about a VLF signal of 1Vpp that you mistakely get on the imput?
Would your speakers like it 20 dB amplified? Or would your crowbar protection nicely trip ?
Carefully set up input capacitor does not make instrumentally measurable distrorsion that I have ever seen in a real measurement setup.
About "sonic" (i.e. audiophools or psycoacoustic) differences, you are naturally intetitled to believe what you want, you may be in a large and respect group, but I am not one of those silver-cable blessers.
effebi
"SILVER CABLE BLESSERS" 🙂
Ahh , good one! Another acronym for "golden ear brigade" or "audiophools". We can not totally condemn these "groups" , for they do by far the most to stimulate the audiophile economy. 😀
A mediocre design with gold, silver ( JFET's), and other esoteric components will be easily outclassed by the superior one made with good ol' lead, tin and copper.
OS
Thanks. But the superior design could be compromised with poor Electros, PVC insulated cables which corrode over time etc., Why not add the best to what is already well designed?
Other than input resistance and input current noise, there is little that JFET's offer over degenerated bipolar input stages. And this is particularly the case when considering input stages to power amplifiers where JFET's offer no significant advantages.
LVF signals would just trip the dc protect.
The nikko amp i had and the telefunken amp i have now are both dc coupled. Not sure if the telefunken uses jfets though, but its dc offset is 0.00000000mV.
The nikko amp i had and the telefunken amp i have now are both dc coupled. Not sure if the telefunken uses jfets though, but its dc offset is 0.00000000mV.
Member
Joined 2009
Paid Member
I've heard sonic differences with output capacitors - never tried different input caps - but I'm more than willing to give the benefit of the doubt to those that have heard sonic differences with different input capacitors. Have you done the experiment yourself ?
We're not talking silver cables here, just differences between penny pinching parts and quality parts.
Not the point
I am speaking reasonable good coupling C against NONE at all.
About having done the experiment, I am a (really) poorly hear gifted engineer, so I must believe what I can instrumentally measure and demonstrate. Having better "sonics" without any number for me is meaningless, I am sorry. And I want to see how many of the so called golden heared converge to the same solution in a blind test. I believe that "sonics" is not only highly subjective, but even not time-invariant 🙂.
Again on topic, i am not saying that FETs are bad, just that they do not bring measurable improvements in practical designs when the subject is a POWER amplifier input stage. Topology (circuit) design and good parametrization (values) seems to play a major role than FET vs BJT.
And then, I quote fully OS - God Bless the Silver Cable Blessers!
I am speaking reasonable good coupling C against NONE at all.
About having done the experiment, I am a (really) poorly hear gifted engineer, so I must believe what I can instrumentally measure and demonstrate. Having better "sonics" without any number for me is meaningless, I am sorry. And I want to see how many of the so called golden heared converge to the same solution in a blind test. I believe that "sonics" is not only highly subjective, but even not time-invariant 🙂.
Again on topic, i am not saying that FETs are bad, just that they do not bring measurable improvements in practical designs when the subject is a POWER amplifier input stage. Topology (circuit) design and good parametrization (values) seems to play a major role than FET vs BJT.
And then, I quote fully OS - God Bless the Silver Cable Blessers!
Anyways where is the best place for the dc offset nulling pot, between the pos/neg rail(depending on if the ltp is npn or pnp) and the current mirror of on the ltp degeneration resistors ?
If you take the time to hfe match the diff pair and have a cap to ground in the feedback path to bring DC gain to 0, there shouldn't be significant DC offset at the output. I haven't found the need for a trimmer.
- Status
- Not open for further replies.