Andrew,
There is no cause for outrage. This is just another audio amp with a slightly compromised biasing schema. Remember, these devices have only recently become available and most designers would not be aware of the slight differences in tempco between output device and internal diode. Another twelve months and everyone will know of this quirk.
Neither is it true that no one listens to you. You are usually fastidious about your interpretations. But in a group of 100 people, only a couple will be really paying attention, this is normal and you should not be disappointed.
Design of any functionally reliable machine is about attention to detail. Despite Mihai's comments, I am convinced that very small emitter resistors increase failure rates, particularly with careless use, which is a fact of life. The bias of a Class AB PP SS amp is tetchy at the best of times, and very small emitter resistors, or worse still, no emitter resistors, are a meltdown waiting to happen.
I would not roundly condemn this amp, either. It is competently designed and the pcb is very pretty. I tend to agree with SandyK. I have not yet heard the sonics, always my principle concern, so can't comment on how it sounds. But the input stage and VAS seems fine to me, although I would always do it differently, but that's just me.
Quoting Mark Twain, "When we remember we are all mad, the mysteries of life disappear and life stands explained."
Cheers,
Hugh
There is no cause for outrage. This is just another audio amp with a slightly compromised biasing schema. Remember, these devices have only recently become available and most designers would not be aware of the slight differences in tempco between output device and internal diode. Another twelve months and everyone will know of this quirk.
Neither is it true that no one listens to you. You are usually fastidious about your interpretations. But in a group of 100 people, only a couple will be really paying attention, this is normal and you should not be disappointed.
Design of any functionally reliable machine is about attention to detail. Despite Mihai's comments, I am convinced that very small emitter resistors increase failure rates, particularly with careless use, which is a fact of life. The bias of a Class AB PP SS amp is tetchy at the best of times, and very small emitter resistors, or worse still, no emitter resistors, are a meltdown waiting to happen.
I would not roundly condemn this amp, either. It is competently designed and the pcb is very pretty. I tend to agree with SandyK. I have not yet heard the sonics, always my principle concern, so can't comment on how it sounds. But the input stage and VAS seems fine to me, although I would always do it differently, but that's just me.
Quoting Mark Twain, "When we remember we are all mad, the mysteries of life disappear and life stands explained."
Cheers,
Hugh
I'll comment in more detail after I see Master Hugh tomorrow - hopefully by then I'll be at a point that I can do a listening test. However in the mean time, here is a story of a litany of my failures which may serve as a road-map for others of where not to travel...:
In other news, I bought a semi-conductor tester from Altronics today, in order to find a couple of well matched pairs for the current mirrors. Thanks to my litany of failures I was well off from spares to select from! Every trannie I tested had a Vbe of exactly 0.82, so it seems that matching this isn't a challenge when using this kit. The Hfe's varied wildly however, so it took a while to find a couple of well matched pairs.
Finally, I should note for any newbies who may be quietly lurking, that spending some time with a master craftsman like Hugh is an extremely valuable exercise. Although one can learn a lot by reading, actually watching an expert at work provides lots of really useful tips and tricks.
- Accidentally reversed polarity from PSU
- Put the Thermal Traks in upside-down - wouldn't fit in the heat-sink then of course, but when I then turned them the right way up I forgot to also reverse their order
- Shorted the pins of a trannie when trying to take a test with my multimeter
In other news, I bought a semi-conductor tester from Altronics today, in order to find a couple of well matched pairs for the current mirrors. Thanks to my litany of failures I was well off from spares to select from! Every trannie I tested had a Vbe of exactly 0.82, so it seems that matching this isn't a challenge when using this kit. The Hfe's varied wildly however, so it took a while to find a couple of well matched pairs.
Finally, I should note for any newbies who may be quietly lurking, that spending some time with a master craftsman like Hugh is an extremely valuable exercise. Although one can learn a lot by reading, actually watching an expert at work provides lots of really useful tips and tricks.
Hi Aksa,
it's not outrage.
I'm just reminding all that the bias circuit is just as important as the rest of the schematic.
Keep in mind that I said
My datasheet is rev2, dated June 2006. How long does a competent designer need to check for stable currents?
it's not outrage.
I'm just reminding all that the bias circuit is just as important as the rest of the schematic.
Keep in mind that I said
just sort the bias and we are ready to listen reliably.
My datasheet is rev2, dated June 2006. How long does a competent designer need to check for stable currents?
Andrew, I just used the values provided by the Peak Atlas DCA55 unit. According to it, Ic of 2.5mA was used to get the Hfe, and Ib of 4.45mA was used to get Vbe.
I'm afraid I'm a complete novice, so these are all just meaningless numbers to me! Does it sound like this process isn't going to provide a good match?
I'm afraid I'm a complete novice, so these are all just meaningless numbers to me! Does it sound like this process isn't going to provide a good match?
Jeremy
What is the Altronics catalogue number of this tester ?
O.K. It is the Peak Atlas DCA55 .
As Andrew has stated, the VBE reading is obviously in error. A typical small signal transistor is more likely to have a VBE of 600-650mV when tested with a typical transistor tester.
Andrew
Jeremy has already stated that he is an absolute beginner in the DIY area,although he is highly respected in an associated field. Perhaps with more guidance from Hugh,he will be in a better position to answer these questions, and either realise the error he made, or find that his new tester is defective.
Greg
Andrew is already in possession of this information.
SandyK
What is the Altronics catalogue number of this tester ?
O.K. It is the Peak Atlas DCA55 .
As Andrew has stated, the VBE reading is obviously in error. A typical small signal transistor is more likely to have a VBE of 600-650mV when tested with a typical transistor tester.
Andrew
Jeremy has already stated that he is an absolute beginner in the DIY area,although he is highly respected in an associated field. Perhaps with more guidance from Hugh,he will be in a better position to answer these questions, and either realise the error he made, or find that his new tester is defective.
Greg
Andrew is already in possession of this information.
SandyK
AndrewT said:
Hi Andrew,
I think you will find "an amp that does not require bias adjustment" is the main driving force in this amplifier design. The resultant kits sell in the thousands and many of the builders would not be able to set the bias properly.
If this bias design gets "close enough" to minimise cross-over distortion and prevent thermal run away then they have meet their design criteria IMHO. Some may not be happy with the solution but hey...
Don't forget these guys have designed heaps of amps over the years and have all the appropriate test equipment and the ability to use it.
regards
Bob Cordell has recently reposted his measuring jig for matching smaller transistors. This can be adapted to measure power devices.
The 200ld idles @ ~ 70mA per device.
This is the Ic current that must be used when checking both hFE and Vbe of each device.
If the 0r22 is substituted for the 0r1 emitter resistor, I would be very tempted to run the same idling current i.e. Vre=7*2.2=15 to 16mV across 0r22, instead of 7mV across 0r1.
Once matched sets with both hFE within 5% and Vbe equal (same Vbe in mV) then set up pairs and find over what range of current you can maintain that same 5% tolerance of Ic, aim for at least 50mA to 150mA.
The 200ld idles @ ~ 70mA per device.
This is the Ic current that must be used when checking both hFE and Vbe of each device.
If the 0r22 is substituted for the 0r1 emitter resistor, I would be very tempted to run the same idling current i.e. Vre=7*2.2=15 to 16mV across 0r22, instead of 7mV across 0r1.
Once matched sets with both hFE within 5% and Vbe equal (same Vbe in mV) then set up pairs and find over what range of current you can maintain that same 5% tolerance of Ic, aim for at least 50mA to 150mA.
I recall various auto manufacturers, including Benz, have produced models with non-adjustable front ends. Most were recalled, revised for both camber and castor adjustment, or swiftly amended on the next model.
I think this situation is 100% analogous...... if a decision were taken to produce an AB PP amp which lacked provision for bias adjustment, then at the least it should have largish emitter resistors, around 0.33R.
Of course, device matching would be mandatory, and would include not just hfe, but also Vbe at the prescribed bias current. I use a jig I designed some years back for this purpose, and routinely match Vbe within 1mV and hfe within 3%. Of course, speed is required since the die heats up during testing, but with a bit of experience you get the general idea.
Hugh
I think this situation is 100% analogous...... if a decision were taken to produce an AB PP amp which lacked provision for bias adjustment, then at the least it should have largish emitter resistors, around 0.33R.
Of course, device matching would be mandatory, and would include not just hfe, but also Vbe at the prescribed bias current. I use a jig I designed some years back for this purpose, and routinely match Vbe within 1mV and hfe within 3%. Of course, speed is required since the die heats up during testing, but with a bit of experience you get the general idea.
Hugh
The NJL's are available separately.
http://www.altronics.com.au/index.asp?search=njl&area=srch&option=and&Submit=SEARCH
You don't need to buy another full kit for spare components.
The other tranny's would also be available separately too.
http://www.altronics.com.au/index.asp?search=njl&area=srch&option=and&Submit=SEARCH
You don't need to buy another full kit for spare components.
The other tranny's would also be available separately too.
Re: SC ULD2
I'm sure the other tweaks suggested in this thread can enhance the design further, and listening tests over the coming days will be most interesting...
The one frustration for those following along with the SC instructions is that in the diagram that shows the target voltages for the emittor resistors (7-10mA), the voltages can not be reached - only about 2mA is achieved. In an earlier post Mickyh confirmed with the SC author that this is an error in the design, and requires a resistor in series with the diodes to fix:sandyK said:
Luckily, it's a very easy fix. And as Alex said, the other problems I've had are entirely my fault - not caused by any deficiencies in the design of the circuit.
I'm sure the other tweaks suggested in this thread can enhance the design further, and listening tests over the coming days will be most interesting...
Re: Re: SC ULD 200W AB Amp
Mihai,
How do you test for this? Doug Self discusses using the change in THD during the power cycle to monitor the effectiveness, but I don’t have an AP system. All I can do is monitor the Re voltage during power, and at the instant of removing the signal, see if the voltage drops initially to the normal quiescent value. It usually drops a little further slightly, presumably due to the heat transfer of the (now hotter than die) heatsink? then rises back to the nominal. I’ve always wondered whether this is sensible. Is it?
Regards,
Brian.
roender said:
Mihai,
How do you test for this? Doug Self discusses using the change in THD during the power cycle to monitor the effectiveness, but I don’t have an AP system. All I can do is monitor the Re voltage during power, and at the instant of removing the signal, see if the voltage drops initially to the normal quiescent value. It usually drops a little further slightly, presumably due to the heat transfer of the (now hotter than die) heatsink? then rises back to the nominal. I’ve always wondered whether this is sensible. Is it?
Regards,
Brian.
jp_howard said:
I am not sure what you are baseing you supposition that these are not thermal traks, NJL's are thermal tracks (reefer to On Semi Website), MJL are not you will find to your dismay that the lazy people who did the web page put up a generic picture which does not refeclt the true nature of the product being offered, Farenll and RS use a side comment to indicate that the picture is indicitive only, but Jaycar and Altronics do not, LAZY. I have purchased NJL's from Altronics and I can assure you they are a five legged Horse.
Alfred
Re: Re: Re: SC ULD 200W AB Amp
Hi Brian,
With the amplifier powered on, heat sink was heated with an electrical radiator at 60-70 degree celsius. The Vre was logged with an RS232 DMM. This was one test, the second one was with signal through amplifier just below clipping level. After 10 minutes the signal was removed and Re was monitored.
If i remember correctly, in both tests delta Vre was under 1.5mV over both Res - 7.5mA bias current. The trend was in the safe side, the OPS is a little bit overcompensated.
Cheers,
Mihai
Cheers,
Mihai
Pingrs said:
Hi Brian,
With the amplifier powered on, heat sink was heated with an electrical radiator at 60-70 degree celsius. The Vre was logged with an RS232 DMM. This was one test, the second one was with signal through amplifier just below clipping level. After 10 minutes the signal was removed and Re was monitored.
If i remember correctly, in both tests delta Vre was under 1.5mV over both Res - 7.5mA bias current. The trend was in the safe side, the OPS is a little bit overcompensated.
Cheers,
Mihai
Cheers,
Mihai
My supposition was based on what the guy at Jaycar told me, and based on the picture on their web site - both which it turns out are completely wrong. So I just wasted over $100 by buying the kits rather than the chips...alfredrofe said:
Alfred also pointed out over email that the trannies I've carefully matched are the wrong ones - it's the 2SA970's that need matching, not the current mirror BC546's. Oh well, easily fixed.
The good news is that I now have 2 working amplifiers. Thanks to Hugh's diagnostic expertise, the earlier problem mentioned of one NJL's voltage going to nearly zero and the other taking all the voltage has been successfully identified as being due to a scratch I created in the heatsink. After filing it away, and using the trimpot in series with the diodes, we now have the expected voltages. Since I only took one PSU to Hugh's we could only test in mono - but early indications are that this unit sounds pretty good.
We did however find the voltages were rather unsteady with 0R1's and even 0R22's. At 0R33 we got steady readings. This may be due to me using a 150mm heatsink - the SC article recommends 200mm, which I hadn't noticed when I bought the heatsinks.
Anyhoo, it's at a point now where I can start some listening tests. After a few hours of listening I'll try making some changes (one at a time, with listening tests), including:
- Bigger heatsink, possibly with lower emittor resistors if they're then stable
- Cap upgrades
- SandyK's mirror mod
- The other mods mentioned in Alfred's earlier summary
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.