Diminishing returns. Can the audible improvement (as small as that is likely to be) outweigh the cost? I don't think so...
I used the LM394 super matched pair in my first Patchwork - waste of money IMO.
The 1815 looks like a winner , mouser even has them ... 2sc1815
yes , ECB ... a third version of the BX
, you could also use the first cascoded version (bx1.3) with the C and B crossed , C with a short piece of 1mm HS tubing. BX 1.2 is cascode-less and also uses ECB , with 50v or less rails, the 1815 would be perfect. (below)
OS
Attachments
Seems that everything Zetex does has high Cob.
You really can't do much better than the MPSA18.
That 2SC1815 has lowish gain.
Last edited:
I'm just being a picky "audiophool" , I'm now down to what I can see on the scope at 1uV. I passed the audible "point of silence" a while ago.
Little things , like turnon "ticks" ... any abnormal noise as the rails collapse. I am just sweating the micro - issues now.If I didn't go modular a while back (5$ a voltage stage) , I might of been satisfied before now.
OS
Input pair does not need a lot of gain (100 would do) ,the current mirror and VAS are where you develop most of your open loop gain. I'm sure the 1815 comes in multiple gain groups as well (O , Y , and GR on datasheet).
OS
SNG001 used the BC's on his BX with excellent results. I have the voltage stages to try them all , it all come down to a performance /cost/ sourcing /obsolescence decision. I'm also sure each "specimen" will perform optimally at a different tail current , another factor in choice (CCS adj.).
OS
I understood that the gain of the input devices only affects output offset and input impedance (the more gain, the better). Trim the overall gain with the emitter resistors.
Are you just looking for something different, to make your amps not look the same as everyone elses??
That only fools the rookies. 
Yes, THAT is expensive. This is primarily why I am manipulating my circuit to use other devices, perhaps ALD1105 matched mosfets from Advanced Linear Devices. Much cheaper, and a better name if you want my opinion.
I will use THAT as a fallback device if it I'm not sucessful.THAT 3X0 series matched transistors are not monolithic. They are made from separate dies that are matched together and then mounted in a common package. This is an advantage I think, but to what degree in audio circuitry I am not sure.
There are 4 THAT 340 in this prototype circuit. In addition the output source balast resistors are these TO-126 power metal film resistors from Caddock at a price of $3.75 each.
Top quality I must say but costly. Funny thing about this amp is that the total cost of all 4 output transistors is ~$4.50.The Caddock resistors I plan to keep using, even at that rediculous cost. Is there any alterative out there with the same quality?
Attachments
You are using those as emitter resistors on your outputs? I use Dale - they are only 65 cents each.
I spend the difference on beer - it makes everything sound better
I understood that the gain of the input devices only affects output offset and input impedance (the more gain, the better). Trim the overall gain with the emitter resistors.
Are you just looking for something different, to make your amps not look the same as everyone elses??
No , not that , I just realized what you stated. Using the MPSA18 made my matching more crucial ("hairy"
). I Decided on the middle of the road SS9014 ("C group" -200+ Hfe). Cascoding the input pair gives "higher everything" (gain , impedance , BW) , leaving just the improved noise factor to consider. I have a 1k -25 turn trimmer for offset , but in the last dozen amps I stuffed , the offset was 2-3mv before adjustment needing just a quarter turn or less to zero. It is better to start out balanced than to alter the gain of one of the tails after the fact.OS
No , not that , I just realized what you stated. Using the MPSA18 made my matching more crucial ("hairy"
As you know, my version of the 'DIY amp that is the same as everyone elses but I call my own' has a cascoded front end. To be honest, in building my 6 channel, I didn't match the MPSA18's at all - too lazy at the time. I have no trimmer for offset either. I measured the offset (of course I can't remember what it was) and I'm sure it was low enough for me to accept it.
I wasn't suggesting that the emitter resistors can be different values. I meant the overall gain of the diff pair can be reduced by increasing the emitter resistors. Better to have too much and throw it away than not enough.
BTW , are those amps still in business ??
I know about LTP degeneration. I was just stating that the CM / VAS degen . is a major factor as well. I have over 110DB LF loop gain with Re = 68 , plenty to "burn up" with NFB/TMC. Offset adj , actually does change beta of the diff. pair, countering that 1% tolerance of the tail resistors or the slight mismatch in device beta , or even slight errata coming from the cascode/CM ! (they should be a close match as well (not critical- but close). If you buy a 100 from the same production run you will find quite a few "close ones".( DMM hfe test.)
OS
The 1815 looks like a winner , mouser even has them ... 2sc1815
yes , ECB ... a third version of the BX
BX 1.2 is cascode-less and also uses ECB , with 50v or less rails, the 1815 would be perfect. (below)
OS
Thanks OS,
I may build a BX1.2 VB to compare against the BX1.3. I noticed that you added C6+R16+R17 on the BX1.6. What is the purpose for these three components.
The 2SC1815 on Tayda are dirt cheap - 0.02 each, so I just bought some as general purpose SSTs: 2SC1815 C1815 TRANSISTOR NPN 50V 0.15A
They also have the 2SK117 for 0.22 each.
The minimum order is only $5.
Cheers, Stanley
Last edited:
Yes, still ticking.
I have an overhaul planned - power supply upgrade mostly.
Totally unnecessary but it should be fun.
"1.6" ??? you mean 1.3 ?? , that is for the cascode divider. with 70V rails , it gives 20v to the LTP to allow for FET's, mpsa18's , ss9014's. with your 40v rails the BC's would work without a cascode , with lower rails R16 at 18k , as suggested , would give 14-16V. The cascode at lower violtages has benefits as well , it isolates the LTP (better PSRR).The beta enhanced VAS has such a high impedance that that the cascode's higher impedance has little or no effect. My simulations show slightly reduced THD WITH a cascode. Many low Vgs Fet's can also be used with a cascode. A cascode IS NOT GOOD on a amp like the goldmund as it has a lower impedance VAS , loading is evident !
A tip.. just remove Q3/4 , jumper the E-C's on the 1.3. I MUST use the cascode for my 75V rails , no choice except for using less than desirable high Vce 1845's (120v).
This adventure of mine is a little unnecessary as well , just for absolute OPS physical symmetry (attachment 1) , to get my BX formally cascoded , TMC'ed and with no jury rigging (as it is now). I will even use oak hardwood blocks with slots to slide and hold the voltage stage to the main current stage. I go fully split PS's this time (Vee2 /Vcc2).
To try out all these input devices / topologies would wear out any other amp on DIYA , so you were right .. I had to design it totally different .. not to impress but to actually DO IT.
OS
Attachments
Last edited:
Actually, the cascoded LTP has same output impedance
as its non cascoded sibling...
The difference is that output is better insulated from input.
Last edited:
We both could be right depending on the circuit it is used in (presence of a CM , using it in a VAS) .. try this search:
OS
Last edited:
2SC1815 from Tayda
Hi all,
I finally received my Christmas present from Tayda and it took about three weeks from the day I placed the order (I actually paid for the parts).
The 2SC1815 looks like Toshiba parts & most of the other parts look good to me. I ordered some lateral-MOSFET and they look like genniue Renesas parts.
Seasons greeting to you all.
Cheers, Stanley
Hi all,
I finally received my Christmas present from Tayda and it took about three weeks from the day I placed the order (I actually paid for the parts).
The 2SC1815 looks like Toshiba parts & most of the other parts look good to me. I ordered some lateral-MOSFET and they look like genniue Renesas parts.
Seasons greeting to you all.
Cheers, Stanley
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.