Both Vunce and I tried SMPS and cap Mx with the Alpha, but with the SMPS supply I had, it seemed to be not happy providing the combined 2.7amps at +/-24v even though rated for 5amps. Also going through the cap Mx with 2.7amps and 4.5v drop was 12w dissipation per channel in the cap Mx (if using MOSFETs). It can work - as I have shown it works on the M2, without a cap Mx but a CRC. So I think maybe get a big powerful 10amp capable SMPS and should be alright.
I run a quick test, through regulated dual 10A 30V lab supply I need to inspect the the diode the bc 547c the ltp vbe of track about 10mV- 20mV of. Need to inspect the diode and KSC1845fta.
I did a stimulation over 900 in a row with relevant spice variables. As Mr. Xr stated put the lid on remeasure.
This on is a somewhat picky. Know wonder it taken months of work for some here. To get stable. Mr. Xr please explain your process to near 0 dc/ac start up cold to warm lid on lid off. You built many, this is your expertise area so share
Hi Tonza,
So you finished building the Alpha and are now testing? The 10mV to 20mV is well within "normal" range for the DC offset and it should not vary greater than that over cold to hot with lid on or off. I did not do anything special - just matched the Hfe on the LTP and have them thermally bonded to each other via shrinktube (no thermal compound paste even). Hugh explained that my 0mV was purely coincidence as the sum of all the voltage drops across the LTP/VAS and CCS and outputs, on average, added up to be zero due to statistical variances washing out. If you are at 20mV, it should remain at 20mV hot or cold if LTP's are thermally bonded together.
You could a different N-channel MOSFET with slight up/down Vgs to "tune" your DC offset but I would leave it alone if you are under 30mV which is nothing on an 8ohm speaker (0.7mW).
Yes i have them bonded i tried them thermally with a resistance wire flat, kapton around to about 60 degres celcius on a breadboard to match them . But i measured on pcb the ksa992 from pins base emitter the off 10-20mV of . Problem is AC out 1.7V not good at all input shorted . I got the dc down to 8mV . I won't connect a loudspeaker you now know why.
I'll try tomorrow . Bed time .
1.7vac is indeed a problem. Please mark up schematic with parts substitutions you may have made and DC voltages at all nodes of the actives for us to help debug. What is your bias current? A closeup photo or two is also very helpful.
Hmmm. A few issues with offset, and AC output at zero signal.
OFFSET:
Offset should be within +/-25mV. Within these limits, all is acceptable.
To put this into perspective, a 100mV offset, a huge figure you almost never see, creates 100/6.8 = 15mA of standing current through the voice coil of a WOOFER (not a mid or tweeter because with a passive crossover both has a cap in the HP filter blocking this current). This 100mV offset dissipates 1.47mW. Since the minimum a voice coil can dissipate is typically 10W, 0.00147W seems trivial - AND IT IS!! Keep the offset within +/-25mV, but do not obsess about getting it down to 1mV. It is not necessary, it is unusual, and it has no effect on sound quality.
Offset is controlled by three factors:
1. Stage Current. This current is passed from the two resistors and pot from the positive rail and filtered by C103, the 1000uF (Refer #111 from XRK on 2 Feb 2018). On the ALPHA it should be 2.05mA. You can measure this on the pot provided at R105. A 100R change on the pot will alter offset around 2.5mV, and raising this resistance brings UP the offset, from -2.5mV to 0mV. This is sole adjustment on this amplifier.
2. Offset is controlled from the long tailed pair (LTP) which seeks to hold the base potential identical for both V101 and V102. The differential action relies upon identical Hfe (beta) for the two transistors and significantly the voltage between base and emitter at the identical collector current, which should be 1.025mA for each transistor.
Therefore since both transistors both require a small base current, typically around 1.025mA/beta, maybe around 7uA assuming 145 of beta for both of them. So, you have to supply this small current, on the left (input, V101) transistor it is supplied from R102, the 22k resistor to ground, and on the right (fb, V102) transistor is should be the same 7uA base current but will be supplied from R113, the series fb resistor f 22k. Note that R102 and R113 are the same 22k resistor value. You can replace this value to 18k, or even 10k, but BOTH RESISTORS MUST BE IDENTICAL.
3. The final adjustment for offset is controlled by the pot, R105, but to be really accurate both LTP transistors, V101 and V102, should be MATCHED for BETA and Vbe at 1mA.
Therefore to work correctly and reveal a magnificent 0mV offset, you have to match the LTP transistor not just for beta, but also for Vbe at 1mA, the prevailing current for each when fully balanced.
This is difficult because matching transistors is expensive and not required. You should simply use two transistors from the batch and hope they are close, then trim the R105 pot to achieve 0mV at the output. And don't be too OCD about this, within +/-25mV is good enough and will NOT affect the sound quality.
As for the 1.7Vac Tonza has on his ALPHA I'm baffled. Are they both the same? How are they measured? Is it all hum/buzz or is it broadband noise? What is the quiescent current on the output stage?
Cheers,
Hugh
OFFSET:
Offset should be within +/-25mV. Within these limits, all is acceptable.
To put this into perspective, a 100mV offset, a huge figure you almost never see, creates 100/6.8 = 15mA of standing current through the voice coil of a WOOFER (not a mid or tweeter because with a passive crossover both has a cap in the HP filter blocking this current). This 100mV offset dissipates 1.47mW. Since the minimum a voice coil can dissipate is typically 10W, 0.00147W seems trivial - AND IT IS!! Keep the offset within +/-25mV, but do not obsess about getting it down to 1mV. It is not necessary, it is unusual, and it has no effect on sound quality.
Offset is controlled by three factors:
1. Stage Current. This current is passed from the two resistors and pot from the positive rail and filtered by C103, the 1000uF (Refer #111 from XRK on 2 Feb 2018). On the ALPHA it should be 2.05mA. You can measure this on the pot provided at R105. A 100R change on the pot will alter offset around 2.5mV, and raising this resistance brings UP the offset, from -2.5mV to 0mV. This is sole adjustment on this amplifier.
2. Offset is controlled from the long tailed pair (LTP) which seeks to hold the base potential identical for both V101 and V102. The differential action relies upon identical Hfe (beta) for the two transistors and significantly the voltage between base and emitter at the identical collector current, which should be 1.025mA for each transistor.
Therefore since both transistors both require a small base current, typically around 1.025mA/beta, maybe around 7uA assuming 145 of beta for both of them. So, you have to supply this small current, on the left (input, V101) transistor it is supplied from R102, the 22k resistor to ground, and on the right (fb, V102) transistor is should be the same 7uA base current but will be supplied from R113, the series fb resistor f 22k. Note that R102 and R113 are the same 22k resistor value. You can replace this value to 18k, or even 10k, but BOTH RESISTORS MUST BE IDENTICAL.
3. The final adjustment for offset is controlled by the pot, R105, but to be really accurate both LTP transistors, V101 and V102, should be MATCHED for BETA and Vbe at 1mA.
Therefore to work correctly and reveal a magnificent 0mV offset, you have to match the LTP transistor not just for beta, but also for Vbe at 1mA, the prevailing current for each when fully balanced.
This is difficult because matching transistors is expensive and not required. You should simply use two transistors from the batch and hope they are close, then trim the R105 pot to achieve 0mV at the output. And don't be too OCD about this, within +/-25mV is good enough and will NOT affect the sound quality.
As for the 1.7Vac Tonza has on his ALPHA I'm baffled. Are they both the same? How are they measured? Is it all hum/buzz or is it broadband noise? What is the quiescent current on the output stage?
Cheers,
Hugh
Last edited:
That is easy i took earlier the ltp ksa992 out for another working amp . Output is really 0.4mV dc max in a typical blameless cfa ltp set at 2.14mA , but here is the fun part signal generator typ 1Khz. sine . Second probe at output oh oh 995hz .
So you're suggestion ltp 1mA with ksa992 come on AKSA , when i took a look at mr.xr 1khz sine i typical behaviour is seen at usully at more difficult load over 4mH 500W bass driver at lower freq. for stability inspections. Even the squere at 1Khz is not anyware good and reported as "stable" .But it score 95/100 . Be real a bad design is what is, now i know so thank you . Perhaps a few seminars more with Ltspice, nice try though!!!
As they say you over 45 the learning curve .........
Hi Tonza,
You are here to find out WHY your amp is not working.
Deriding the design, and me with it, doesn't help your problem at all........
When you have gotten over blaming me and X, and admit you have made a simple problem someplace, we will make progress. Give us a few pictures too, you have been asked twice now.
Hugh
You are here to find out WHY your amp is not working.
Deriding the design, and me with it, doesn't help your problem at all........
When you have gotten over blaming me and X, and admit you have made a simple problem someplace, we will make progress. Give us a few pictures too, you have been asked twice now.
Hugh
Tonza,
As I recall, you were going to substitute parts all over the place, including the "hard to find" KSA992, but now you say you are using 992's? What about the 1845, or 547, didn't you switch all of them? What parts are left from the original spec?
Alpha 20 is a really simple amp to build and run. If yours is putting out 1.7VAC, something is wrong with either a part or operator error in assembly. You are the only one having problems with this amp, and if you are not interested in debugging it with our help by providing some measurement of dc voltages at node points and maybe a photo as asked, well, there is not much else to say.
Last edited:
I at work, some of as have to. Other obligations too. I merely gave you money for your enthusiasm.
You haven't shown any voltages and you want mine, I seen this kind of action before , and pics for a none functional circuit. I won't contribute pics. As much as you have not just pics here and there but not one shows any evidence that you are measuring this circuit.
To be clear ones again
NOT ONE OF YOUR PICS SHOWN THERE'S NO EVIDENCE YOUR MEASURING THIS PCB.
1 built as you shown the fafafaa.
I use esd matt with fets just a tip for free and please ground yourself.
I can't publish copyrighted material so contact Jake Rothman at everyday practical electronics and discuss this further as he published 10nf1k bye.
boss will be mad at me for typing at work.
Sorry Hugh and X but i can't help myself.
Tonza 75 If I remember correctly when i posted my response as to the Quality of the ALPHA 20 you replied that you have wasted 4 years at university studying electrical engineering and what is my secret.
WELL if a dumb **** like me with NO formal training can build the ALPHA 20 without any issues or the 20 or so other amps which i have built in the past 5 years also without any Magic Smoke!!!.
THE SECRET is to follow the given instructions.
Tonza 75 If I remember correctly when i posted my response as to the Quality of the ALPHA 20 you replied that you have wasted 4 years at university studying electrical engineering and what is my secret.
WELL if a dumb **** like me with NO formal training can build the ALPHA 20 without any issues or the 20 or so other amps which i have built in the past 5 years also without any Magic Smoke!!!.
THE SECRET is to follow the given instructions.
tonza, turning this into a slanging match will not further your cause. You have some highly respected members on here trying to help you resolve the issues with your amp and as such the least you can do is go along with their suggestions of help. Remember they are all 'volunteers' and under no obligation to post here.
It might also help if post some scope shots of what you are seeing. In particular of this where you seem to be saying a 1kHz input is transformed into a 995Hz output.
