i am working on a simple class AB amplifier and the LTP is formed with a pair of BC 547 ....
i am confused now since in my stock i have Siemens transistors that seem that come from very good batch and all of them measure around 230
i also have Philips that measure from 480 to 620 but it will be easy to locate 10 that measure around 520
i have also ST that measure around 400
what to use ???
why select A against B or C ???
any ideas ?
Kind regards sakis
i am confused now since in my stock i have Siemens transistors that seem that come from very good batch and all of them measure around 230
i also have Philips that measure from 480 to 620 but it will be easy to locate 10 that measure around 520
i have also ST that measure around 400
what to use ???
why select A against B or C ???
any ideas ?
Kind regards sakis
Usually higher hfe is preferable for an input stage.
Unclassified BC547's can be from around 100 to 800 hfe. The Philips ones sound like "C" versions, around 400 to 800 hfe. I would use those.
Unclassified BC547's can be from around 100 to 800 hfe. The Philips ones sound like "C" versions, around 400 to 800 hfe. I would use those.
ok then ..thanks for hopping in
why though what is the benefit of using something with 500 instead one other that is 200? will it sound better ? will it be more linear ? will it be faster ?
is there audible difference ? can i produce one board mad with A transistors and a second board made with B transistors and ry o listen to the difference ?
kind regards
sakis
why though what is the benefit of using something with 500 instead one other that is 200? will it sound better ? will it be more linear ? will it be faster ?
is there audible difference ? can i produce one board mad with A transistors and a second board made with B transistors and ry o listen to the difference ?
kind regards
sakis
One other point:
Lower Hfe -> gives higher Base current -> gives higher input offset ==>gives higher output offset when a cap is used in the feedback path.
Greetz
Lower Hfe -> gives higher Base current -> gives higher input offset ==>gives higher output offset when a cap is used in the feedback path.
Greetz
Chance is, that with higher hfe, you will get a higher Open Loop Gain.
This will give you a better effect of NFB, so you will have less distorsion.
The other side is that you risk more on the stability side.
Generally, also for noise purposes is better to have higher gains in the first stages. So, yes, the difference could be audible to a competent listener.
This will give you a better effect of NFB, so you will have less distorsion.
The other side is that you risk more on the stability side.
Generally, also for noise purposes is better to have higher gains in the first stages. So, yes, the difference could be audible to a competent listener.
ok ...got it ..thank you very much
will make tests for both cases and place my results here
Kind regards
Sakis
will make tests for both cases and place my results here
Kind regards
Sakis
...well i have been preparing a proper setup and willing to see how it will go ...through my Philips stock i found hfe 630 and plenty so i closely matched a few to go ahead .
here is another one ....
LTP is supposed to be balanced all the way through input resistor from base to ground of example of 27k should be equal to feedback resistor also = 27K transistors closely matched and with thermal junction and also current flow from upper side to lower side should also as close is possible or equal
Question is that one may choose to run his LTP at 0.5 ma ,1 ma, 5ma , or 10, ma for example ....what will be the benefit of running an LTP on 0.5 ma or running at 8 ma
I actually don't seek for fast food electronics solutions since i find this very insulting to forum members and since my theory is poor and also don't know any printed or electronic source to trace this information i placed my question here to get information from people that already bounced in this question and come up with a solution .
Thankful regards
sakis
here is another one ....
LTP is supposed to be balanced all the way through input resistor from base to ground of example of 27k should be equal to feedback resistor also = 27K transistors closely matched and with thermal junction and also current flow from upper side to lower side should also as close is possible or equal
Question is that one may choose to run his LTP at 0.5 ma ,1 ma, 5ma , or 10, ma for example ....what will be the benefit of running an LTP on 0.5 ma or running at 8 ma
I actually don't seek for fast food electronics solutions since i find this very insulting to forum members and since my theory is poor and also don't know any printed or electronic source to trace this information i placed my question here to get information from people that already bounced in this question and come up with a solution .
Thankful regards
sakis
What I know is that the higher the current the higher the noise. (current noise)
But with lower current you have lesser current gain bandwidth product. Look at the data sheet. The current source I use delivers 2mA, so that each draw 1mA in stable state.
It is a trade of between noise and performance.
Hope this helps.
But with lower current you have lesser current gain bandwidth product. Look at the data sheet. The current source I use delivers 2mA, so that each draw 1mA in stable state.
It is a trade of between noise and performance.
Hope this helps.
listened to the version with hfe 630
stability wise had no serious effect offset in channel A was 6mv while in channel B was 7.2 obviously My LTP is well balanced at this point there is no trimmer for that and also i don't see any use for it in such ratings
next to previous transistors with hfe of 260 there is a bit more punch in the bass ...bass remains also very tight thanks to close matching and thermal junction middle gained a click i think in presence while crystal clear and very transparent.
high i think is also one more click cleaner while informative detailed and plenty ...
so far so good
thank you
stability wise had no serious effect offset in channel A was 6mv while in channel B was 7.2 obviously My LTP is well balanced at this point there is no trimmer for that and also i don't see any use for it in such ratings
next to previous transistors with hfe of 260 there is a bit more punch in the bass ...bass remains also very tight thanks to close matching and thermal junction middle gained a click i think in presence while crystal clear and very transparent.
high i think is also one more click cleaner while informative detailed and plenty ...
so far so good
thank you
Sakis these are interesting findings which I correlate with and theory explains and shows it well. That said I was told by some a while back on this forum that no sound differences could be heard between different input transistors.
These are some of the remarks regarding theory I posted before on this topic on the forum and the main reasons why their are audible effects :
In LTP s (high beta) they reduce input bias currents and this has several positive effects. Here is a list of advantages :
Reduction of input offset voltage.
Reduction in input current distortion when input is fed from higher source resistances as usually happens with long cable runs, volume controls and preamp / sources that dont have very low output impedances.
Higher ripple rejection.
Try mpsa18 for further improvement.
ok ..all this is turning to an interesting issue ....
next question will be operating voltages
hardman cardon very often operates semis on the edge .... on the edge means very close to max voltage operation and much close to max current operation .... it was mentioned also quite often that practice like that ""measure"" better but will produce no serious or audible difference
in a few words operating an LTP that is formed as above IE : balance techniques /hfe choice / and rails of 35-35 volts a BC547 with max operation voltage of 50V ( for example ) will work ... but also a MPSA 92 ( for example ) will also work but the MPSA is a 250 volt device if i remember well why use A against B ?
any comments on that ?
-----------------------------------------------------------------------------------
given the opportunity often i noticed people asking questions regarding rail capacitors where rail voltage was too close or marginal next to the operation voltage of the capacitor
seen commercial amplifiers working in the proper voltage no 220-240 issue ( amp was rated to 240 volts ) where the mains was 230 volt and rail voltage was 63 volts and the filter capacitors also rated at 63 volt
since there was short of debate and some said that all caps should have a margin for over voltage i actually placed the question to the Nippon Chemicon division of Athens
The replay was : the best operation condition for a capacitor and the best performance will be when the operation voltage is the same as stated on the capacitor
Additionally what is written on the capacitor is the operating voltage and not the maximum operation voltage .
To blow a 63 volt capacitor will require at least 90 volt and for a period of time ...
that was the Nippon Chemicon
next question will be operating voltages
hardman cardon very often operates semis on the edge .... on the edge means very close to max voltage operation and much close to max current operation .... it was mentioned also quite often that practice like that ""measure"" better but will produce no serious or audible difference
in a few words operating an LTP that is formed as above IE : balance techniques /hfe choice / and rails of 35-35 volts a BC547 with max operation voltage of 50V ( for example ) will work ... but also a MPSA 92 ( for example ) will also work but the MPSA is a 250 volt device if i remember well why use A against B ?
any comments on that ?
-----------------------------------------------------------------------------------
given the opportunity often i noticed people asking questions regarding rail capacitors where rail voltage was too close or marginal next to the operation voltage of the capacitor
seen commercial amplifiers working in the proper voltage no 220-240 issue ( amp was rated to 240 volts ) where the mains was 230 volt and rail voltage was 63 volts and the filter capacitors also rated at 63 volt
since there was short of debate and some said that all caps should have a margin for over voltage i actually placed the question to the Nippon Chemicon division of Athens
The replay was : the best operation condition for a capacitor and the best performance will be when the operation voltage is the same as stated on the capacitor
Additionally what is written on the capacitor is the operating voltage and not the maximum operation voltage .
To blow a 63 volt capacitor will require at least 90 volt and for a period of time ...
that was the Nippon Chemicon
Last edited:
I'm not to good on this "noise" issue, but I think the current noise only becomes a problem when the source resistance/impedance becomes higher.
For sources where Rs~1k0 and often <200r current noise is not an issue.
For low voltage noise generally, input transistors perform better with a high current bias.
For very low source impedance, very high bias currents are generally quieter.
I have seen BJT LTP running with a tail current of 12mA and a cascoded version even higher.
jFET LTP also seem to benefit from high bias of similar values, both alone and cascoded for lowest noise with low Rs.
For sources where Rs~1k0 and often <200r current noise is not an issue.
For low voltage noise generally, input transistors perform better with a high current bias.
For very low source impedance, very high bias currents are generally quieter.
I have seen BJT LTP running with a tail current of 12mA and a cascoded version even higher.
jFET LTP also seem to benefit from high bias of similar values, both alone and cascoded for lowest noise with low Rs.
Last edited:
What about transistor noise figure. There is a lot info to be read ...
For an example BC550 has batter NF than 547.....
Regards
For an example BC550 has batter NF than 547.....
Regards
Transistor geometry dictate that high breakdown voltage transistors have much lower betas as is the case with MPSA92 so youd lose all the advantages mentioned. I doubt youd be able to find a MPSA92 that has half the beta of BC547c. Another important factor to look at is Cob that will contribute to high frequency distortion, the less the better, here the bc is better too I suspect.
Operating close to max voltage will minimize Cob which is good for high frequency THD, you can see the effect on the datasheet with Vce - cob.
Last edited:
AGAIN ALL THIS IS VERY HELPFUL !!!
been listening for about 8hours as exactly said :replaced BC546 with BC 547 run them in 35+35 rails and select ones with hfe of 630 ...
its superb
been listening for about 8hours as exactly said :replaced BC546 with BC 547 run them in 35+35 rails and select ones with hfe of 630 ...
its superb
A couple things I know: The Ic needs to be in the flat part of the hfe/Ic curve. The current is what drives the next stage (usually the compensation cap of the voltage stage or the input capacitance of the transistor)). Higher current means a higher gm for the input stage, which usually helps with stability. Higher Ic also means a higher base current, which can increase your offset. I think there are other effects just dont remember right now.
Yes it minimizes Cob but it will increase the Early effect ( Ic change with Vce ) which can be a bigger problem, depends on the circuit, this effect can be eliminated with cascodes.
Higher current does mean higher gm but in contrary that it helps with stabilty it makes things worse, you must up the degeneration untill you get the same gm or youll have to up the compensation. But not all is bad if you do this as higher current is then available to charge compensation cap and slew rate and high frequency linearity is improved which is very good. The downside is the increase in noise but then again there are ways to improve on this.
The higher base current which causes THD with non zero source impedance is exactly the reason you should use transistors with high betas, when using larger LTP biasing current this becomes of even greater importance.
I see it as the lesser of two evils, where as the early voltage affects the lower frequencies THD (1Khz down) where its fairly simple to have low THD, the Cob affects the high frequencies (1Khz and up) where its difficult to improve and is normally much higher and easy to pick up by the ear. In a LTP situation early voltage has nearly no effect, the beta does especially with high source material which is nearly always the case and the Cob is important for low high frequency THD.
Cascoding is the better way to go although you can make things worse if not used properly but very good results and in many cases better can be had without it if the transistors Vceo permit.
Keep in mind Im only refering to LTP situations on this.
- Status
- Not open for further replies.