C5 and 9 can both be 35V. C5 is the CCS decoupled reference , it see's
about 1/3 rail voltage (20V+ for 60V rails). Try for the largest value (not critical) at 35/50V.
C9 is the Vbe decoupler , it see's just 4 X Vbe (2.4V-2.7V depending on bias).
16/25/35V is usable, the value can be 10-22u (not really critical) ...
it should be a quality cap (panasonic low esr).
OS
Your Mouser saved project only had general purpose Panasonic 22uF ( EEA-GA1V220H Panasonic | Mouser ). I hope that they will be OK. I will look through what I have on hand in case I happen to have any low ESR caps in that value range.
OS, what do you mean by the largest value? What would be some sensible limit? 220uF? 1000uF?
Thank you,
cheers,
Largest uF that will fit the new V2.4 board
. It has 5MM ! pads for C5/9 !Looks like even a 220u/63V C5 will fit the new pads. I would stay at
100u/ 63V max.
Too large a cap might increase "thump" (Higher T R/C might delay CCS stabilization).
One way to offset this drawback is to have lower R in the divider (R11-13).
This will allow more current ( Q8/12) .
I am using this on the "wolverine" (6.8K-6.8K-100uF). The only effect this
has is to reduce the minor saturation of the VAS CCS during a clipping event.
That is also why I split my reference up for the wolverine , a clip event
at the VAS won't affect the LTP CCS's reference.
Unless you are going to use the badger as a guitar amp , these changes
are overkill.
Also , a little extra PSRR will be gained by changing C5 from 47u --> 100u.
My best recommendation would be C5=100u/50-63V (5mm pitch) and
decrease R13 to 10K.
OS
Largest uF that will fit the new V2.4 board
Looks like even a 220u/63V C5 will fit the new pads. I would stay at
100u/ 63V max.
...
Also , a little extra PSRR will be gained by changing C5 from 47u --> 100u.
My best recommendation would be C5=100u/50-63V (5mm pitch) and
decrease R13 to 10K.
OS
Now you are talking about upgrading 47 to 100uF... I have the lower end of the BOM recommendation with only 22uF... Is this going to be detrimental? In my case, I don't really want to tack an additional cap behind the board since that would put the cap between the board and the heatsink which would keep it on the warm side. I also only have 10uF caps to add.
I also still only have a general purpose 22uF Panasonic for C9 instead of a low ESR. Is this going to be less than ideal? I don't want to order again if it doesn't really matter. I can order again but being in Canada, it's $20 shipping each time. I already had to order twice because my DMM died on me.
I didn't know that this shared Mouser cart was trying so hard to pinch pennies... that was not my intent with this build. This amp is already costing me $750+ , a few extra dollars for quality caps would be a drop in the barrel.
I HFE matched all the transistors that you said should be matched (with my DMM at least), I had extra of every small transistor so was able to match perfectly at least within the limitations of my meter. I matched the resistors on the LTP and the others that you mentioned... I think it was R20 and R21. I chose the highest HFE 992 and 1381 for q9 and q10... I have 2 extra pairs of drivers so I will try to match them as much as possible, and a couple extra output pairs so I can use the ones that match the closest. Should I be matching Vbe or HFE on the drivers and the output transistors?
Since I do also have extra transistors for q13, what should I be looking for here? (ie: does vbe or hfe matter much?)
If I wind up ordering again, I could replace all the electrolytics with low ESR versions, (except the Bi-polar which seems is not possible) is there anything else I should be considering when ordering to make improvements?
Thanks for any help on these matters. I know I am worrying about small details, but I certainly don't want any compromises that are only saving me a couple dollars.
The 22u will work ... 47-100u are overkill !
But ,most small caps are cheap - why not ?
Some blameless designs have NO cap , and still work.
In my text guide (that the build guide is based on) , I explained the extreme
range of sourcing that would get a working amp.
I know some like to augment/embellish. When I made the Mouser BOM for myself ,
I had little money .... so I kept the BOM at 50-60$.
Then offered it to the forum for convenience.
The text BOM suggests 3-4 semi choices for each stage of the amp ,
as well as ranges /types for the passives. We (and I) wanted to make this "global" ,
(not everyone has mouser/digikey).
Ask about any part and you shall receive a concise explanation.
I never got around to update the BOM ... I thought of a "economy" / deluxe
dual BOM - never got to it.
For Q13 .... the 3503c is 40-80Hfe , the Vbe circuit it is in is an EF2.
EF2's are not very "picky" if they have good layout and an adequate
heatsink. BD136-140 can also be used.
BOTTOM line ... Most general purpose semi's can be used with VERY
little change in performance."Badger" caps and resistors are so derated that
there are no stressed parts.
Audiophile Input caps and high quality power supply components are
the best way to spend "extra money". Your careful matching and selections
will pay off when you see just a couple of millivolt offset and
play /use this project until your children have to re-cap it 20+ years from now.
OS
But ,most small caps are cheap - why not ?
Some blameless designs have NO cap , and still work.
In my text guide (that the build guide is based on) , I explained the extreme
range of sourcing that would get a working amp.
I know some like to augment/embellish. When I made the Mouser BOM for myself ,
I had little money .... so I kept the BOM at 50-60$.
Then offered it to the forum for convenience.
The text BOM suggests 3-4 semi choices for each stage of the amp ,
as well as ranges /types for the passives. We (and I) wanted to make this "global" ,
(not everyone has mouser/digikey).
Ask about any part and you shall receive a concise explanation.
I never got around to update the BOM ... I thought of a "economy" / deluxe
dual BOM - never got to it.
For Q13 .... the 3503c is 40-80Hfe , the Vbe circuit it is in is an EF2.
EF2's are not very "picky" if they have good layout and an adequate
heatsink. BD136-140 can also be used.
BOTTOM line ... Most general purpose semi's can be used with VERY
little change in performance."Badger" caps and resistors are so derated that
there are no stressed parts.
Audiophile Input caps and high quality power supply components are
the best way to spend "extra money". Your careful matching and selections
will pay off when you see just a couple of millivolt offset and
play /use this project until your children have to re-cap it 20+ years from now.
OS
Last edited:
merlinx76,
You may buy resistors and capacitors on EBay quite safely and in most cases shipping is free. Just look at vendors' ratings for particular items you intend to buy. For power amps I buy 1W 1% metalised resistors (20-25pcs sets) and Dale RN60 1/4W 1% or better where low noise is needed as standard. Most through hole resistors and caps are made in China anyway. 1W metalised resistors are bigger but have lower noise chracteristics than the so called 0.6W which are hardly 1/4W.
Alternative to Mouser is Michael Percy who specialises in supplying quality audio parts to DIY community: Michael Percy Audio Ordering Information Check his catalogue for prices and shipping costs.
cheers,
You may buy resistors and capacitors on EBay quite safely and in most cases shipping is free. Just look at vendors' ratings for particular items you intend to buy. For power amps I buy 1W 1% metalised resistors (20-25pcs sets) and Dale RN60 1/4W 1% or better where low noise is needed as standard. Most through hole resistors and caps are made in China anyway. 1W metalised resistors are bigger but have lower noise chracteristics than the so called 0.6W which are hardly 1/4W.
Alternative to Mouser is Michael Percy who specialises in supplying quality audio parts to DIY community: Michael Percy Audio Ordering Information Check his catalogue for prices and shipping costs.
cheers,
Everywhere I look, opinions are all over the place about whether matching is worth worrying about for drivers. It seems that with the 15030/31 and 15032/33 they are hopelessly mismatched with huge variations in one transistor vs its comp. pair.
I see in the Symasym thread that the MJE15034/35 seem to be a much closer matched set over varying conditions. They are not really any more expensive that the 15032/33. Would they possibly be a better choice for drivers?
I see in the Symasym thread that the MJE15034/35 seem to be a much closer matched set over varying conditions. They are not really any more expensive that the 15032/33. Would they possibly be a better choice for drivers?
It's probably unlikely that I could hear a difference from this one issue however they are off by more than a typical amount with the hfe of one being about double the complementary version (at least on my DMM). I may just order some 15034/35 and use the opportunity improve a couple of my capacitor choices while I am ordering.
It seems the little details are holding me up anyway so there is no worry about waiting for parts since I still have things to keep me busy (Drilling /tapping/ learning!)
It seems the little details are holding me up anyway so there is no worry about waiting for parts since I still have things to keep me busy (Drilling /tapping/ learning!)
mje15034-35 may be used without any problems in HB.
mje 15xxx transistors and power transistors should be matched at higher Ic currents than available in multimeters. ON data sheet does not show hfe of these transistors below 100mA, while all multimeters measure hfe at about Ic=2mA and some at even lower currents. Check in your DMM manual Ic current used in Hfe measurements. If Ic is about 2mA or below I would not worry much about this mismatch. I must say that my mje15xxxs are pretty close even at Ic=2.5mA - one is not much blow 100 the other not much above 100.
Generally matching transistors makes sense where matching visibly (sonically) affects performance or is necessary because of amps design and that has to be done at required operational currents. So differential pairs have to be matched and mirror pairs have to be matched. To some extent it also pays to match power transistors. In some cases high Hfe is required - here Q9 so I'm using 2sa970bl with Hfe of 460 at 2.5mA.
cheers,
mje 15xxx transistors and power transistors should be matched at higher Ic currents than available in multimeters. ON data sheet does not show hfe of these transistors below 100mA, while all multimeters measure hfe at about Ic=2mA and some at even lower currents. Check in your DMM manual Ic current used in Hfe measurements. If Ic is about 2mA or below I would not worry much about this mismatch. I must say that my mje15xxxs are pretty close even at Ic=2.5mA - one is not much blow 100 the other not much above 100.
Generally matching transistors makes sense where matching visibly (sonically) affects performance or is necessary because of amps design and that has to be done at required operational currents. So differential pairs have to be matched and mirror pairs have to be matched. To some extent it also pays to match power transistors. In some cases high Hfe is required - here Q9 so I'm using 2sa970bl with Hfe of 460 at 2.5mA.
cheers,
Last edited:
matching hFE of the drivers is not terribly important in a 3 stage or triple EF output stage.
In a two stage EF the VAS to CCS current passes into and back out of the driver bases.
A slight error in hFE does not matter.
BUT !!
when the hFE is doubled, it means the base current is halved.
The sucking and blowing does not match.
The CCS is sending a set current down the line. But the drivers are removing varying amounts. The VAS at the end of the line sees a very varying current and thus cannot perform properly.
Try to get the hFEs within say 20%, or adopt a pre-driver topology.
In a two stage EF the VAS to CCS current passes into and back out of the driver bases.
A slight error in hFE does not matter.
BUT !!
when the hFE is doubled, it means the base current is halved.
The sucking and blowing does not match.
The CCS is sending a set current down the line. But the drivers are removing varying amounts. The VAS at the end of the line sees a very varying current and thus cannot perform properly.
Try to get the hFEs within say 20%, or adopt a pre-driver topology.
If the driver beta is of a concern , just run the VAS a little "hotter" (R27=56).
PS - cut a larger heatsink out of .020 or flashing(aluminum) , it can be
longer than normal ... all the way from "D-bc-option" (the clamp diode) back
to the positive rail. 75mm X 75mm should dissipate the 11mA+" hot" VAS.
Edit - ran simulation of a totally mismatched driver pair . NJW0302 was used for the PNP , mje15032
for the NPN. THD20 rose from 8 ppm to 16ppm with 8ma VAS .... from 7 to 11ppm for 11ma VAS.
The "hotter VAS" does reduce the effects of the mismatch. Bias pot has to be lower with a
11ma vas , but still well within range.
OS
PS - cut a larger heatsink out of .020 or flashing(aluminum) , it can be
longer than normal ... all the way from "D-bc-option" (the clamp diode) back
to the positive rail. 75mm X 75mm should dissipate the 11mA+" hot" VAS.
Edit - ran simulation of a totally mismatched driver pair . NJW0302 was used for the PNP , mje15032
for the NPN. THD20 rose from 8 ppm to 16ppm with 8ma VAS .... from 7 to 11ppm for 11ma VAS.
The "hotter VAS" does reduce the effects of the mismatch. Bias pot has to be lower with a
11ma vas , but still well within range.
OS
Last edited: