The purpose of this new thread is to both explore an obscure voltage amp (the Koda), and to discuss the REAL WAY to implement 15 pairs of modern BJT's to get that holy POWER 😛 😀 !!
First with the amplifier , I have the koda 261 schematic , made by a Hong Hong company..
I never heard of them before peeking at their amp. I am glad I did , as the Voltage amp varient I have implemented below is truely astonding in it's harmonic byproducts , gain ,ultra low THD, and most of all ... it's utter SIMPLICITY. 🙂
Below is the schematic of the clone (along with a 1000 watt / 4R triple OPS). It works by integrating the typical CCS in a standard brainless (blameless) amp into the emitter follower of the main VAS. The operating point of the CCS (KSC3503) is set by a simple 2 diode string. This is in turn modulated by the current through the 100k resistor (r16) and the MPSA42's emitter.
I did not think this would work with any precision when I first saw it. Boy ... I was wrong . It has 10db more gain than the typical blameless and -6 - -10db better in the THD department. The harmonic products of the FX are inline with the BX (bootstrap) or the CX (symasym) , just 10 X lower distortion.
On to the triple OPS with explanations. I truly question why anyone would want a full KW for home use 😕😕 . The only use I could envision is driving a 21" german made PULSE subwoofer or other bass unit
for a club or disco. In this case using an voltage stage with ultra low distortion or a better harmonic structure is an exercise in futility. Just using a standard .1% thd RCA/DX stage ported to a higher rail voltage would be adequate.
A 500 watt amp , at least in the headroom department , will lead to a feeling of effortlessness with a better sound stage. That is why , at least for this amp , am going to stick to my 4 pair of OP devices. The devices chosen... the NJW0281/0302, are REAL CHEAP , in N. America they are $1.14 apiece in quantities of 10 or more . SO ... 15 pair (30 = $34) are WAY cheaper than 10 pair MJL (20 X 4 = $80). They are VERY good devices at 150W / 250 Vceo /70Hfe . Being smaller you can also fit 15 pair in the same physical space as 10 of the typical OP's.
As far as the triple OPS , it is the only way to make this many crude non-linear devices in parallel do what you want them to do ( BIG AC arc welder with variable line frequency). For 1Kw @4R , you MUST at least be aware of Mr. SOA (as charlie calls it). Below you see the NJW's do about 700ma @ 100V and 1A at 80V. At 15 pair , each device will pass 500mA at 500W/8R .. well within the DC SOA @ 80V. At 1000W /4R , You would be pushing the SOA limits a > 1A per device. With typical Disco use (Boom Boom) , you would be relying on the 100 Ms SOA curve (2A @ 80V) , so all would be good ... as long as you had sufficient heatsinking.
The amp below I believe would run flawlessly as it has an ideal bode plot for loop gain stability. The triple (my greatest worry) could also be degenerated further at the pre-drivers lowering the overall gain of the triple. As it stands the triples current gain is 40 X 40 X 50 = 80,000 (best case) ... kind of on the lower end for the gain of a triple, more like the typical darlington. This is very good for the Voltage amp , as it only sees 70- 80 uA load .. allowing it to have .0002% @ 1khz and .0001%(last pix below) at 100Hz (typical use for a 1KW amp !)
P.S, LTspice for 1KW FX1.1 is below.
OS
An externally hosted image should be here but it was not working when we last tested it.
First with the amplifier , I have the koda 261 schematic , made by a Hong Hong company..
I never heard of them before peeking at their amp. I am glad I did , as the Voltage amp varient I have implemented below is truely astonding in it's harmonic byproducts , gain ,ultra low THD, and most of all ... it's utter SIMPLICITY. 🙂
Below is the schematic of the clone (along with a 1000 watt / 4R triple OPS). It works by integrating the typical CCS in a standard brainless (blameless) amp into the emitter follower of the main VAS. The operating point of the CCS (KSC3503) is set by a simple 2 diode string. This is in turn modulated by the current through the 100k resistor (r16) and the MPSA42's emitter.
I did not think this would work with any precision when I first saw it. Boy ... I was wrong . It has 10db more gain than the typical blameless and -6 - -10db better in the THD department. The harmonic products of the FX are inline with the BX (bootstrap) or the CX (symasym) , just 10 X lower distortion.
On to the triple OPS with explanations. I truly question why anyone would want a full KW for home use 😕😕 . The only use I could envision is driving a 21" german made PULSE subwoofer or other bass unit
An externally hosted image should be here but it was not working when we last tested it.
for a club or disco. In this case using an voltage stage with ultra low distortion or a better harmonic structure is an exercise in futility. Just using a standard .1% thd RCA/DX stage ported to a higher rail voltage would be adequate.
A 500 watt amp , at least in the headroom department , will lead to a feeling of effortlessness with a better sound stage. That is why , at least for this amp , am going to stick to my 4 pair of OP devices. The devices chosen... the NJW0281/0302, are REAL CHEAP , in N. America they are $1.14 apiece in quantities of 10 or more . SO ... 15 pair (30 = $34) are WAY cheaper than 10 pair MJL (20 X 4 = $80). They are VERY good devices at 150W / 250 Vceo /70Hfe . Being smaller you can also fit 15 pair in the same physical space as 10 of the typical OP's.
As far as the triple OPS , it is the only way to make this many crude non-linear devices in parallel do what you want them to do ( BIG AC arc welder with variable line frequency). For 1Kw @4R , you MUST at least be aware of Mr. SOA (as charlie calls it). Below you see the NJW's do about 700ma @ 100V and 1A at 80V. At 15 pair , each device will pass 500mA at 500W/8R .. well within the DC SOA @ 80V. At 1000W /4R , You would be pushing the SOA limits a > 1A per device. With typical Disco use (Boom Boom) , you would be relying on the 100 Ms SOA curve (2A @ 80V) , so all would be good ... as long as you had sufficient heatsinking.
The amp below I believe would run flawlessly as it has an ideal bode plot for loop gain stability. The triple (my greatest worry) could also be degenerated further at the pre-drivers lowering the overall gain of the triple. As it stands the triples current gain is 40 X 40 X 50 = 80,000 (best case) ... kind of on the lower end for the gain of a triple, more like the typical darlington. This is very good for the Voltage amp , as it only sees 70- 80 uA load .. allowing it to have .0002% @ 1khz and .0001%(last pix below) at 100Hz (typical use for a 1KW amp !)
P.S, LTspice for 1KW FX1.1 is below.
OS
Attachments
Last edited:
was there i thingy called VI limmiter in the original amplifier that is no longer there any more ????
and by the way ....why a bell rings that koda has diferent voltage supply in the front end than in the power stage ??? ( i write this before looking at the schematic but normally i trust my kingstone )
common guys !!!! amplifier at this rate of power has to stop being an LTP ,vas, drivers, and output .... at this point you are doing the same thing LIke carlos did ....design a good amp and skip all the rest ....
and by the way ....why a bell rings that koda has diferent voltage supply in the front end than in the power stage ??? ( i write this before looking at the schematic but normally i trust my kingstone )
common guys !!!! amplifier at this rate of power has to stop being an LTP ,vas, drivers, and output .... at this point you are doing the same thing LIke carlos did ....design a good amp and skip all the rest ....
Last edited:
Os, its not really new, the vas is called a super pair, theres a lot of discussions here on it, its quite difficult to stabilize, likes to oscilate in the Mhz range. Yamaha has also used it in a lot of their amps especially in subwoofer amps. Theres a thread here called high-loop gain amp where another member also shows some sims using this setup except he went further and enhanced it even more, it has ultra low distortion, the question is if one can successfully build it and how it would sound.
I know it is not new !!! nothing is around here. I am glad to know what it is really called. You are wrong about it liking to oscillate. NO ,it will work without a CDOM... rare! with 56P and some tweaking it is nicer than a brainless.
Sakis , As far as that Reeeeeeediculous 1000 watt amp , you can buy a class D 1KW for $300 , so why bother DIY. AND OF COURSE I would add a VI limiter to it. The original is below (koda261) and has no VI , its bigger brother(koda266) is a 400 watter with a VI limiter ... I have no schema of it, but have a collection of 23 VI limiters I could use with it!! 😛
The 261 uses just 1 PS , 55-0-55v, the 266 has a separate winding on its toriod I would guess 70-0-70 and something else (maybe higher --- 80-0-80??)
Back to the "Super pair" amp. Below is the cheap chinese iteration of it. Also below is my proposed build (FX1.1VB) attached to my PB250 current board.
Much better (even cleaner) than with the triple above. What I did notice is that it DOES get unstable if you use a higher gain tranny for the input (Q7 in schema below).
This is a popular amp with soundcard fanatics and their "souped up" modded
auzentech's. It is said to have a VERY clean sound and there are whole threads devoted to modding the 261.
I really like the super pair !!! 🙂🙂
Edit I am wrong.... the 261 does have a crude VI , even for 100w.
OS
Attachments
Last edited:
that could be very interesting and trully i would like to see the 23 circuits of VI limmiters that you have since i am only aware of one or two
may be a vi limiter thread ??? i woyld be very hapopy to see some inovation there
kind regards sakis
may be a vi limiter thread ??? i woyld be very hapopy to see some inovation there
kind regards sakis
Found the "super pair" http://www.diyaudio.com/forums/solid-state/25172-baxandall-super-pair.html
In that thread , I see it was said that as an emitter follower , the super pair was very stable. This is the very way it is in the Koda. 🙂 NO CDOM .WOW!!
Could not find where someone was "tweaking it". Looking at the real amps schematic , the truth of the pair not liking capacitive loads is apparent in Koda's choice of the main compensation at the input pair.
I tried the cap B-C on the KSA1381 (Q8) and B-C on the input tranny (Q1) with no change whatsoever. There must be a good reason why Koda placed it on Q1 , so I won't argue.
Below is the REAL sim of this thing as I will make the artwork for. I have the real amp before me (sounds great for being a cheap corporate production) , so there is no mystery in making a 200w version of it. 🙂 Really does not matter as I will soon be able to go from what I see to what I hear in a single day (modular rules)
OS
In that thread , I see it was said that as an emitter follower , the super pair was very stable. This is the very way it is in the Koda. 🙂 NO CDOM .WOW!!
Could not find where someone was "tweaking it". Looking at the real amps schematic , the truth of the pair not liking capacitive loads is apparent in Koda's choice of the main compensation at the input pair.
I tried the cap B-C on the KSA1381 (Q8) and B-C on the input tranny (Q1) with no change whatsoever. There must be a good reason why Koda placed it on Q1 , so I won't argue.
Below is the REAL sim of this thing as I will make the artwork for. I have the real amp before me (sounds great for being a cheap corporate production) , so there is no mystery in making a 200w version of it. 🙂 Really does not matter as I will soon be able to go from what I see to what I hear in a single day (modular rules)
OS
Last edited:
That is a good idea, since everyone hates VI limiters I can practice being a pitchfork villager and pounce on them.
Here is the first one(koda261). I actually have 60+ but they are contained in my 1k amp schematics.
OS
Attachments
Os what do you mean by cleaner than with the triple, find anything wrong with a triple setup, but I dont mean with a sim, the real thing.
Do you know if 2sa2223 and 2sc6145 are widely available there and from which supplier ???
Ive just run your sim, at around 1Mhz I see some oscillation, I have tried build a amp or two with this type vas hoping that the sim was a bit wrong but real time scope also showed some problems. Im still a bit in the dark about it as Yamaha have so many amps like this and for sure they wouldnt sell something that could cause problems.
Do you know if 2sa2223 and 2sc6145 are widely available there and from which supplier ???
Ive just run your sim, at around 1Mhz I see some oscillation, I have tried build a amp or two with this type vas hoping that the sim was a bit wrong but real time scope also showed some problems. Im still a bit in the dark about it as Yamaha have so many amps like this and for sure they wouldnt sell something that could cause problems.
Last edited:
There is much discussion on the Baxandall super pair in the Forum.
I wonder if Q7 & Q8 need relative selection for speed (fT) and gain (hFE) and for relative Ic of both devices to keep away from the instabilities that are alleged?
I wonder if Q7 & Q8 need relative selection for speed (fT) and gain (hFE) and for relative Ic of both devices to keep away from the instabilities that are alleged?
Your curves are innaccurate in the sense that what is effectively
measured in term of bandwith is dominated by the input filter 820R/220pF.
Remove this caps for accurate bandwith and phase sims....
The closed loop response look then less rosy....
Last edited:
Andrew it could well be, I tried sims on some of the yamaha amps using this vas, they all display oscillation, apexaudio a member here also has a amp here that he uses the super pair, his amp oscilates so badly on sim its scary yet he builds them. Maybe one can get away with a amp oscilating in the Mhz range ???
For one , I have the KODA261 sitting in front of me , along with my $20 fleabay hitachi 20mhz scope .... NO 1mhz oscillation. To respond to Wahib's comment - the ac sweep runs on the NFB loop .. input cap makes no difference . I did sweep the input and produced your readings exactly. THANKS
Andrew ,right on the money ... the super pair did oscillate when I used a Hfe300 ksc3503 model as Q7. Sub that with a lower gain (60-80) MPSA42 , everthing is good. BTW , I balanced this amps LTP and went from .0015% to .0002% @ 1khz / 30v p/p .. Any slight slight instability in this amp is ONLY evident with almost no emitter degeneration at the input pair or the mirror. At this point the amp had almost 120db OLG !
I then solved that with a little more lead compensation (15p) fed back from the VAS. I then did a 20KHZ transient test with a 32v p/p square wave ... perfect , slightly rounded edges , VERY good slew. Below are the plots. 🙂
OS
Attachments
Someone previously posted a modified (enhanced) schematic of the DSE-A2760 (which they state is the same as a Koda261). It also contains a baxandall 'super pair' VAS : http://www.diyaudio.com/forums/soli...760-amplifier-aka-koda-261-a.html#post1835800
Since this has been built, and presumably tested properly, we must assume that this super pair VAS works (or can be made to work).
With my own sims (attached), a ZTX694B and an MJE15035 is also stable, although I do use a 100pF miller cap round the whole pair, in order to define the open loop gain.
Since this has been built, and presumably tested properly, we must assume that this super pair VAS works (or can be made to work).
With my own sims (attached), a ZTX694B and an MJE15035 is also stable, although I do use a 100pF miller cap round the whole pair, in order to define the open loop gain.
Attachments
thanks , Ihan .. I could not find this with a "super pair' or "high open loop gain" search.
I notice my sims - 1khz @50w are in line with his (the thread) at 1W !
I never thought to wrap the miller around the whole pair
it does bring the gain down to a normal brainless amps level ( 70+DB)😛 .OS
Attachments
Do you have the time (or inclination) to build it? If you do, I'd be interested in the stability of the Baxandall super pair VAS.
It is just another of the modules that will hook into the PB250 (below), they only cost $10 apiece. I've started it (the FX1.1 - koda) ..already (2nd pix - template).
If I have too many problems I just unplug it , plug in something else 😎.
OS
Attachments
works great , had to increase the current source (VR - R6 :all my CCS's are trimmable) to 3.2ma my VAS ended up at 7mA , which means I had to rebias the OPS back to 70mA per device.
In the end , same THD , A little less loop gain (you could change R11-12 to 47R to compensate). Small ,easily made "tweaks". This is best done on the real thing while "scoped" 🙂
OS
Q90 and 91 should be larger devices IMO considering thermal derating, SOA, idle current, and survival (hot) with a short circuited output.
I'd probably use one per output if you must use that device at that supply voltage.
Have you tested into 1 or 2 ohm loads, or are you not allowing those. Lower supply V would be suggested for low Z loads.
I'd probably use one per output if you must use that device at that supply voltage.
Have you tested into 1 or 2 ohm loads, or are you not allowing those. Lower supply V would be suggested for low Z loads.
Last edited:
- Status
- Not open for further replies.