Yes and pretty sure he post when ready. Here at thread look post #1660, other thread see link http://www.diyaudio.com/forums/solid-state/239708-vssa-50-a-3.html#post3579500 and look forward. Also seen at PMI's CM thread he is in progress and plan to use CM PSU for the setup.
You right, and pretty sure it is minor because have experinced 4 times to break those SANKEN output devices thruu years (Not fun for speaker with + or - 63V DC, if it gets + the cone looks like a fisheye from deepsea fish family). Had the first of these amps in 1983, played live music with them lots of times and wondered then why they broke down. Here later and wiser i see not smart to hot plug in/out musical instruments with mixer master pot turned up on a real DC coupled amp meant for HiFi. Anyway can turn it on, but says some strange very small buzzing which i suspect to be the feedforward devices 2SC2238/SA968 one or both, but lets see. First time it broke in late 80's i got pro repair, it did cost a fortune and the techman repaired twice, because he was not used to DC coupling at that time, so when he turned on without checked all devices for error, he broke the just replaced ones too. Later i got little hobby knowledge + schematic and repaired myself.
That can't be bad or hurt anything. I have not finish these trials but am in the middle of it, and direction seems to be that that i can form/manage the edge of sound picture (envelope/groupdelay). But most fasinating is when going as linear in impedance as you can with good mastered recordings you get a step further into the studio and things can be seperated more. This votes in line with one at the thread saying he gets soft good musical treble out of a YAMAHA tweeter he earlyer did know as not so good maybe cheep sounding.
Link earlyer this thread give some hint http://www.diyaudio.com/forums/vend...-lateral-mosfet-amplifier-67.html#post3409577. Am sure theres more answers somewhere here in thread, and pretty sure of 80W 8R/120W 4R with +/-45V DC.No input about 2x25 300VA sizing ?
Last edited:
Thanks dear Esperado !
80w is more than what I was expecting... great. Such a transformer (2x25v 300VA) will provide 10A, perfect to feed 2 VSSA module on 8ohms, if we consider that each module will draw 35/8 = 4,35A... but still on the small side considering 4ohms speaker, if I want to get full dynamic headroom : is that right ?
80w is more than what I was expecting... great. Such a transformer (2x25v 300VA) will provide 10A, perfect to feed 2 VSSA module on 8ohms, if we consider that each module will draw 35/8 = 4,35A... but still on the small side considering 4ohms speaker, if I want to get full dynamic headroom : is that right ?
Last edited:
In fact, the max peak to peak level in simulations is 58V (+-29V). It makes ~52 Watts eff under 8 Ohms. But L.C. said, if i remember well, 80watts. and 140watts for 4 ohms.
Remember, in your calculs that quiescent current will be ~170mA, witch add ~13Watts.
With 4 ohms, sims said 52V (+- 26V) witch provide 84 Watts eff and will eat +- 6.5 Amps for each module.
My shoot (if Esperado or Andrej pop up better trust them).
Behind my thinking i assuming that first two device poles are checked and surrounding resistors all OK, else check these. Then priority:
1 - J1/J2 should be zero ohm, expect them to be more (Exchange with 0 ohm 0,5 W SMD 1206 (3216)).
2 - Use diode test for D1,D2,D3,D4 (Exchange with TZMC 15 V, 0,5 W, MiniMELF SOD-80, Vishay).
3 - If point1/2 OK then it must be TLVH431 (Exchange with TLVH431DQ SOT-23-3, Texas Instruments).
It will be diffycult to exchange any of those bits without removing at least one of 1mF/50V at pos or neg side. Component values are taken from the early stage BOM #post705, and best schematic at post #300, and good luck.
1/2/3 are all fine. In a rush of desperation did indeed exchange the tl431, yet obviously it was doing exactly what it was supposed to - provide 2.8 - 3.1v. After replacing got the voltage of the spreader even down to 2.5v by reducing R17 and it obviously cannot get less than that.
I may be doing something profoundly stupid and not realising it but at this stage it looks like the replacement output device is just more sensitive. Is this possible at all?
I may be doing something profoundly stupid and not realising it but at this stage it looks like the replacement output device is just more sensitive. Is this possible at all?
Check R5/R6=1K R21/R22=20R has direct lane to ALF device. You have OK module to do point compare measure. If sure all resistors and points measure equal between defect/OK module then ALF defect Again. Do all power checks with 22R 5W on both rails.
About ALF device here is text from manual page.....
Warning: ALFOBNPl6Vd5 dual mosfet transistor
is a static electricity sensitive device and does
not include internal Gate-Source protection
network, so great care must therefore be taken
with static electricity handling precautions.
Just measured the transfer characteristic of the ALF. The desired bias current of 168mA is reached at +0.793v at the gate (tested only the N-part). So, i guess twice this is required for the two, or about 1.58v between the two gates. It seems this is where i get confused: how is this achieved with a TL431?
The current sources fix the quiescent current in the input stage. This quiescent current create a DC voltage across their collector resistances. This voltage makes the VAS tranies conducting proportionally.(~10mA) because it is more than their base/emitter diode voltage. This current in the VAS create a voltage across the common resistance between their collector, witch is more than the minimum voltage required by the power mosfets gates to conduct.
This collector resistance is a variable one,
That is why a 1.8mA current in the input stage create a ~10mA in the VAS witch create a quiescent of ~150 mA in the output stage that you can adjust with the VAS collector's pot..
As-it, the input stage has a 1 gain factor because the impedances of the current sources + the emitter resistances are higher than their collector resistance.. But L.C. used capacitors to shunt the impedance of the current sources to 0 for AC signals. Now, the input stage is, for AC signals, just a common emitter stage with a gain set by the collector/emitter resistances ratio.
May-be your problem is you setlled the tuning pots the wrong sides ?
The VAS collector resistance is a variable one, and has to be fixed at the minimum value before to fire your amp for the first time. As-it you will have no current in the output stage. Once you have verified that there is 10mA in the VAS, adjusting the Current sources, you can begin to tune this pot, in order to have a quiescent in the output, then fine tune the Current source to minimize the DC offset in the output: read carefully L.C. instructions.
Anyway, you have to protect your amp, before tuned, with little resistances added in serial to the power supply rails. Like this, if too much current in the output stage, they will blow-up, it is less expensive than power MOS FETS.
When everything is nicely tuned, remove those resistances, verify all currents ok, let the amp take his temperature, then fine tune the output quiescent current if necessary: your amp is ready to use
This collector resistance is a variable one,
That is why a 1.8mA current in the input stage create a ~10mA in the VAS witch create a quiescent of ~150 mA in the output stage that you can adjust with the VAS collector's pot..
As-it, the input stage has a 1 gain factor because the impedances of the current sources + the emitter resistances are higher than their collector resistance.. But L.C. used capacitors to shunt the impedance of the current sources to 0 for AC signals. Now, the input stage is, for AC signals, just a common emitter stage with a gain set by the collector/emitter resistances ratio.
May-be your problem is you setlled the tuning pots the wrong sides ?
The VAS collector resistance is a variable one, and has to be fixed at the minimum value before to fire your amp for the first time. As-it you will have no current in the output stage. Once you have verified that there is 10mA in the VAS, adjusting the Current sources, you can begin to tune this pot, in order to have a quiescent in the output, then fine tune the Current source to minimize the DC offset in the output: read carefully L.C. instructions.
Anyway, you have to protect your amp, before tuned, with little resistances added in serial to the power supply rails. Like this, if too much current in the output stage, they will blow-up, it is less expensive than power MOS FETS.
When everything is nicely tuned, remove those resistances, verify all currents ok, let the amp take his temperature, then fine tune the output quiescent current if necessary: your amp is ready to use
Last edited:
Apparently we are not looking at the same circuit. There is no "common resistance between their collector", instead there is a TL431 spreader. The rev1.4 is close enough representation of the actual circuit and this is what is under discussion here - the commercial kit and not some earlier revision.
In the case of a variable resistor one can easily set bias to zero by choosing zero resistance. Not so with a TL431. But i have already described the issue in my post above.
The common resistor Esperado mention is TR3 (variable) sitting between collector of VAS pair and accross/paralell to TLVH431, is ours TR3 blown off the PCB when first ALF leaved its life ??
Not sure if it's a language thing, or i just cannot describe anything to save my life, so here is a more concise attempt :
1. Lazy Cat board working fine - bias of VAS and outputs is correct, offset is miniscule
2. Something, not part of this discussion, occurs and leaves a blown output as well as a couple of blown protection zeners. All other parts on the board appear fine.
3. New output is soldered in place. It biases much higher.
4. Looking at the component values on my board it appears the bias voltage can vary between 2.8 - 3.1v with the pot which is only a 10% variation. Using the tl431, bias cannot be made under 2.5v anyway. There is also a good chance i cannot make head or tails from the circuit in which case could someone please set me straight?
5. I desolder output device and test half of it on the bench. At 0.793v at the gate it conducts 168mA through the drain.
Could it be new ALF, being obviously from a different batch has a significantly different transfer characteristic?
:1. Lazy Cat board working fine - bias of VAS and outputs is correct, offset is miniscule
2. Something, not part of this discussion, occurs and leaves a blown output as well as a couple of blown protection zeners. All other parts on the board appear fine.
3. New output is soldered in place. It biases much higher.
4. Looking at the component values on my board it appears the bias voltage can vary between 2.8 - 3.1v with the pot which is only a 10% variation. Using the tl431, bias cannot be made under 2.5v anyway. There is also a good chance i cannot make head or tails from the circuit in which case could someone please set me straight?
5. I desolder output device and test half of it on the bench. At 0.793v at the gate it conducts 168mA through the drain.
Could it be new ALF, being obviously from a different batch has a significantly different transfer characteristic?
Last edited:
I just had a look to the schematic. On my sims, with laterals 2Sk1058 & 2SJ162, the required gate to gate voltage seems to be around 2.4V.
Indeed, tl431 has a minimum of 2.5V. Out of range ? Better ask L.C or somebody with a working amp, as i don't know the exact behavior of the ALF. But something is strange on your parts, 0.793v is damn low (the N and P parts are supposed to be different).
Strange, it works for me. Direct link
Edit: Post #1062http://www.diyaudio.com/forums/vend...lateral-mosfet-amplifier-107.html#post3476745
Last edited: