ours sound like the 2*5*2*0?
- Thread starter Mbira
- Start date
Help Support GroupDIY Audio Forum:
deveng
Well-known member
Fabio,
I've done quite a few Pspice simulations on that circuit and I agree with the 10pf cap change. It should improve the high frequency response. The 47pf value sim shows a rather nasty dip in the upper end of the bode plot. It still is not ideal though. I found that increasing the tail current on the input differential pair smooths the response to almost ideal. Changing the 22k value to about 10-12k is enough to do the job. I havent done any breadboarding of this yet but intend to do so. If anyone would like to try this change I'd be interested in the real world results. I'm aware that simulations don't always translate perfectly to the real world but I believe these changes are not simulation only justified. Someone with the design skills and intuition (like PRR) may want to comment on this.
regards,
Jeff
Hey Jeff,
Great you are doing this. I did 2 blocks first, with 82k and 20k on the input pair, they sound almost identical to me at first hearing. Now I need to go deep and make some measurements. A friend are getting good results (soundwise) with 20k there. Will try your 12-10k and see!
Thanks for the tip. There´s so many options there on the schematic...
Yea, it would be great to have more comments on this.
:guinness:
Fabio
Great you are doing this. I did 2 blocks first, with 82k and 20k on the input pair, they sound almost identical to me at first hearing. Now I need to go deep and make some measurements. A friend are getting good results (soundwise) with 20k there. Will try your 12-10k and see!
Thanks for the tip. There´s so many options there on the schematic...
Yea, it would be great to have more comments on this.
:guinness:
Fabio
Flatpicker
Well-known member
[quote author="deveng"]I found that increasing the tail current on the input differential pair smooths the response to almost ideal. Changing the 22k value to about 10-12k is enough to do the job...[/quote]Are we talking about R1 here? All the schematics I have show it as 150K. :? Or are you refering to the 25K used from emitter to V- when using another npn as a current source?
deveng
Well-known member
Yes that's right. I've only Pspiced a few of these 2*5*2*0 circuits and I was not particularly interested in the versions without the current source so I didn't do too much with those. I believe you could go up to as much as 100 to 120ua on the 'combined' tail currents (50-60ua's on each of the input transistor emitters). This would translate to around an 8k resistor in place of the 25k.
I just Pspiced Peter's 1731 schematic today and so far it looks good right out of the box. It has a 130k emitter resistor (no NPN current source) and it shows about 110ua. A 150k drops it to about 96ua.
regards,
Jeff
peter purpose
Well-known member
My last disection had 6k85 hanging off the emiter and a diode measuring 1231 hanging off the base.
peter
peter
deveng
Well-known member
I've been playing around with the tail current values on the 1731 and found some interesting things happening. First, at lower gains typically used on mic preamps with condensor mics (10-20db) the Pspice plots show a peaking at 1.5Mhz. According to Doug Self's discrete op-amp design webpage: http://www.dself.dsl.pipex.com/ampins/discrete/discrop.htm
He is referring to the ratio of value of the resistors on the collectors to the tail values. If I reduce the 1731 schematic tail resistor value to about 20k the bode plot smoothes out very nicely. I see a similiar thing happening on the 2*5*2*0 circuits when the current set by the current source is too low. Doing a bit of research it appears that small signal transconductance is proportional to tail current making the circuit non linearly voltage dependent. I'd like to see someone with PRR's design experience and intuition jump into this one.
If some of the schematics floating around don't have these values set properly, it may explain why they may not sound as good as they can (or closer to the real thing). Bench testing with the proper equipment may reveal some even more interesting things.
Comments anyone....
regards,
Jeff
He is referring to the ratio of value of the resistors on the collectors to the tail values. If I reduce the 1731 schematic tail resistor value to about 20k the bode plot smoothes out very nicely. I see a similiar thing happening on the 2*5*2*0 circuits when the current set by the current source is too low. Doing a bit of research it appears that small signal transconductance is proportional to tail current making the circuit non linearly voltage dependent. I'd like to see someone with PRR's design experience and intuition jump into this one.
If some of the schematics floating around don't have these values set properly, it may explain why they may not sound as good as they can (or closer to the real thing). Bench testing with the proper equipment may reveal some even more interesting things.
Comments anyone....
regards,
Jeff
Flatpicker
Well-known member
[quote author="deveng"]Yes that's right... [/quote]Ok, I'm with you now. BTW thanks for all of this interesting info! :thumb:
Peter, I´m lost here! :?: :!: :?:
I´m playing mainly with R3 and R4... did one with 82k and one with 20k.
Now in the schem there´s a lot of GONE parts... lots of untested possibilities... :roll:
TAMAS STEP IN PLS!!!
cheers!
peter purpose
Well-known member
[quote author="Bauman"]Peter, I´m lost here! :?: :!: :?: [/quote]
Ciao Fabio,
The last one I pulled apart had the following values....
R3 49K5
R4 8K26
R5 39K5
R2 390R
Q10 emitter R18 6K85
Q10 base R17 is a diode
peter
Ciao Fabio,
The last one I pulled apart had the following values....
R3 49K5
R4 8K26
R5 39K5
R2 390R
Q10 emitter R18 6K85
Q10 base R17 is a diode
peter
deveng
Well-known member
I think we need to be careful here that we're not talking about different schematics. I'm new to this forum and at this moment (after a long recording session and it's late evening here) I don't know how to post a schematic! So, I'll try to clarify any confusion I may have added here! On most of the 2*5*2*0 schematics I've seen Q1 has a resistor from the collector to +15v (R3?) and 2 resistors from Q2 collector to +15v. The 2 resistors on Q2 collector (R4 and R5) were for the 'trim' pin to be taken from between them to adjust for offset. Since this is not so necessary when using coupling caps and transformers we could just combine these 2 to a single value and not use the trim pin at all. The tail current is set by the resistor on the emitter of the current source transistor (Q10).
My point was that the tail current resistor should be adjusted to increase the tail current to about 100-120ua. Then the collector resistors should be changed to be approximatley 2 to 2.5 times this value. Also, if the collector resistors are changed so should the cap that is parallel with the resistor tied to Q1 collector.
This is where a simulator comes in to play (for those of us who don't like math). Pspice can be used to determine a "good" value for all these components. Many of my circuits that were spiced first have translated nicely to real circuits and performed well. I strongly urge any DIY'r to get a decent simulation program and experiment. A lot can be learned with this tool.
I'll try to post some schematics and maybe even some sim plots to demonstrate my points. A bit busy these days but I'll give it a shot.
Regards,
Jeff
My point was that the tail current resistor should be adjusted to increase the tail current to about 100-120ua. Then the collector resistors should be changed to be approximatley 2 to 2.5 times this value. Also, if the collector resistors are changed so should the cap that is parallel with the resistor tied to Q1 collector.
This is where a simulator comes in to play (for those of us who don't like math). Pspice can be used to determine a "good" value for all these components. Many of my circuits that were spiced first have translated nicely to real circuits and performed well. I strongly urge any DIY'r to get a decent simulation program and experiment. A lot can be learned with this tool.
I'll try to post some schematics and maybe even some sim plots to demonstrate my points. A bit busy these days but I'll give it a shot.
Regards,
Jeff
So yet more options :? Oh man...
What I see is that, again, we have a difference between R3 (49k5) and R4+R5 (8k26+39k5). Aren't they suposed to be equal? Looks like the answer is NO, on Peter report there's a difference of 1k74 and on the ones I have it's 8k (R3=82k and R4+R5 = 90k). The trim leg is have more resistance on this last case.
On the other hand on NOTES it suggest R3=20k and R4+R5= 19k9 (3k9+16k), which makes the current on both legs the same...
What I can tell you is that it works with 20k on both sides... but why the difference ?!
:? All about the input pair transistor?
:guinness:
Fabio
deveng
Well-known member
Collector current = emitter current (essentially) so a bit of difference in the collector resistor values should only result in a slight difference in the voltage drops. The key is still the tail current setup either by a single resistor from Q1 & Q2 emitters to -15v or in the case of a current source, the resistor from the current source emitter to -15v. I wouldn't worry to much about the collector resistors but focus more on the current source and it's effect on linearity, transconductance and overall frequency response.
If the collector resistors are 20k, I'd try a 7.5k resistor for the current source emitter value and a 3k for the resistor tied to the base of the current source and -15v. This should give you around 120ua tail current. Also use a 62pf cap in parallel with the Q1 collector resistor ( not the trim leg).
On versions without the current source, a little Pspice work may be needed to determine the optimum value for the tail current.
Regards,
Jeff
Sleeper
Well-known member
Does anyone have pics of the 2520r268 boards assembled?
Just about to etch a bunch of these sice I already have some bc550 and bc560s left over from my green pres.
the drawings and artwork all make it pretty clear, but a pic of the real thing just helps loads on the assembly...
:sam: :sam: :grin:
Just about to etch a bunch of these sice I already have some bc550 and bc560s left over from my green pres.
the drawings and artwork all make it pretty clear, but a pic of the real thing just helps loads on the assembly...
:sam: :sam: :grin:
I found an error on the 2520r268 board parts layout. The Q3's base and emitter are mixed. They should be the other way around. I've built one and now it's working well without oscillations.
Thanks for reverse engineering this :thumb: .
Okko
Thanks for reverse engineering this :thumb: .
Okko
I used a 47pf np0 ceramic.
I have done only a quick test and it seems to be working. I haven't made any measurements yet.
Here are some pictures of the 2520r268 and some other op-amps I've made so far (and one extra).
From left: JE990 (with bc550/560 and bd139/140 transistors), Forssell class a jfet, Melcor M1731, 2520BC, QE AM10 and 2520r268.
After listening some music through them I found out that the melcor, and the two apis sound quite similar. Melcor is a little softer and the 2520BC has a little more bass than the 2520r268 in a quick listening test.
The AM 10 is my favorite in a line op application. It has a great character but it's quite clean at the same time.
The JE990 and mr Forssel's jfet are both very clean but slightly different from each other.
These tests were nothing too serious, just the first impressions so don't make any conclusions based on this.
Now I have to make mic pre cards to make some more testing. I'll post more impressions after I'm done. Is anyone interested in sound clips? I could make some during the tests.
Cheers!
-Okko
Flatpicker
Well-known member
[quote author="omeiko"]...Is anyone interested in sound clips?[/quote]Yes!
