Sexy Triode Killer ?
Congratulations Robert on your success, and thankyou for the good information.
This is very encouraging that you are now getting this level of performance.
Have you tried with the input coupling transformer again ?.
270W/4ohms, 300W/2ohms sounds very usefull.
Did you try 2-3 ohms for extended period to gauge heating/reliability ?
The data sheet shows seperated connections for input stage and output stage - this is maybe a usefull feature also - maybe regulated supply for the input stage and switchable supply voltages for the output stage only to give a degree of protection when loaded with less than 8 ohms ?.
This technique is used in a number of Jap amplifiers and shelf systems.
I am still weighing up which approach that I will use for a multiway/multiamp DJ hire system, so your results are most usefull to me.
I like the idea of cheap modules that easily 'plug-in' appeals to me from performance, reliability and servicability angles very strongly !.
Regards, Eric.
Congratulations Robert on your success, and thankyou for the good information.
This is very encouraging that you are now getting this level of performance.
Have you tried with the input coupling transformer again ?.
270W/4ohms, 300W/2ohms sounds very usefull.
Did you try 2-3 ohms for extended period to gauge heating/reliability ?
The data sheet shows seperated connections for input stage and output stage - this is maybe a usefull feature also - maybe regulated supply for the input stage and switchable supply voltages for the output stage only to give a degree of protection when loaded with less than 8 ohms ?.
This technique is used in a number of Jap amplifiers and shelf systems.
I am still weighing up which approach that I will use for a multiway/multiamp DJ hire system, so your results are most usefull to me.
I like the idea of cheap modules that easily 'plug-in' appeals to me from performance, reliability and servicability angles very strongly !.
Regards, Eric.
STK vs LM3886
hi robert,
that is pretty impressive that an IC part could rival your Rowland. i was about to build a simple compact integrated for my dad based on the National LM3886 chip but the STK stuff is starting to look more interesting... my primary reason for sticking with the LM3886 is the simplicity, the external component count is almost nothing even compared to the STK. you wouldn't have tried the LM3886 have you? i'm curious how it compares to the STK in sound quality... no doubt the STK has far greater power and drive capability but at low to moderate volume on, say, small 2-way monitors, i wonder which sounds better. in any case i'll probably still build the LM3886 amp, my twist on it will be to use a very slim, compact chassis (1 RU high, 8" deep) and choke regulation for the supply. power will be limited (probably around 35WPC) but i'm just trying to make a simple little integrated for a second system.
cheers,
marc
hi robert,
that is pretty impressive that an IC part could rival your Rowland. i was about to build a simple compact integrated for my dad based on the National LM3886 chip but the STK stuff is starting to look more interesting... my primary reason for sticking with the LM3886 is the simplicity, the external component count is almost nothing even compared to the STK. you wouldn't have tried the LM3886 have you? i'm curious how it compares to the STK in sound quality... no doubt the STK has far greater power and drive capability but at low to moderate volume on, say, small 2-way monitors, i wonder which sounds better. in any case i'll probably still build the LM3886 amp, my twist on it will be to use a very slim, compact chassis (1 RU high, 8" deep) and choke regulation for the supply. power will be limited (probably around 35WPC) but i'm just trying to make a simple little integrated for a second system.
cheers,
marc
eric,
1. I ran the 2 ohm load just for a minute or two in order to get read-outs at clipping. The heat sinks are not designed for that continuous of a load. (To tell the truth, the chassis is a gutted Quad 405 II. The only original part is the chassis itself, the transformer and the bolted on rectifier. I re-tapped the front heat sink panel on the inside to bolt the STK modules onto it. It is a nice little package and far, far better than the original.) I did not re-use the input transformers.
2. The other reason I ran it for a short while is that the STK modules have no inherent current limiting nor muting features. So destruction of the output devices is possible if one is sufficiently determined.
Marc,
3. This leads to your answer: since there are no internal extras, the sonics from the STK may be better than the LM3886. This is also one reason why I compared it to the Rowland. The Rowland model 10 amp is a parallel-bridged LM3886. I would think of it as the culmination of what is possible with this device. One other difference is that the LM3886 is quasi-complementary and the STK modules of the 'XI' ilk are fully complementary.
4. Which leads to one correction I need to post from my first on this thread: the STK4048 uses 4 output devices not 6 and the STK4044 2 devices not 4.
5. A friend came over yesterday and we did another Rowland/STK comparison. He too heard the soundstage difference but otherwise thought them strikingly similar in tonality.
6. I have pair of Le Amp's that I now plan on using for a bridged trial using a pair of STK4044XI at +/-36 VDC with maybe larger caps on the input stage section of the STKs. The power supply in these monoblocks is quite good and I believe the STK sounds better than the TDA7294. I'm not sure of the exact power output when derating from 50 V to 35 V or so. Probably it will be good for around 80 W and 4x that bridged. I have to ponder this one over.
Regards, Robert
1. I ran the 2 ohm load just for a minute or two in order to get read-outs at clipping. The heat sinks are not designed for that continuous of a load. (To tell the truth, the chassis is a gutted Quad 405 II. The only original part is the chassis itself, the transformer and the bolted on rectifier. I re-tapped the front heat sink panel on the inside to bolt the STK modules onto it. It is a nice little package and far, far better than the original.) I did not re-use the input transformers.
2. The other reason I ran it for a short while is that the STK modules have no inherent current limiting nor muting features. So destruction of the output devices is possible if one is sufficiently determined.
Marc,
3. This leads to your answer: since there are no internal extras, the sonics from the STK may be better than the LM3886. This is also one reason why I compared it to the Rowland. The Rowland model 10 amp is a parallel-bridged LM3886. I would think of it as the culmination of what is possible with this device. One other difference is that the LM3886 is quasi-complementary and the STK modules of the 'XI' ilk are fully complementary.
4. Which leads to one correction I need to post from my first on this thread: the STK4048 uses 4 output devices not 6 and the STK4044 2 devices not 4.
5. A friend came over yesterday and we did another Rowland/STK comparison. He too heard the soundstage difference but otherwise thought them strikingly similar in tonality.
6. I have pair of Le Amp's that I now plan on using for a bridged trial using a pair of STK4044XI at +/-36 VDC with maybe larger caps on the input stage section of the STKs. The power supply in these monoblocks is quite good and I believe the STK sounds better than the TDA7294. I'm not sure of the exact power output when derating from 50 V to 35 V or so. Probably it will be good for around 80 W and 4x that bridged. I have to ponder this one over.
Regards, Robert
Giant Killer
Hi Robert,
That 'Rowland model 10 amp' is pretty well regarded isn't it ?
The bridge comparison test sounds interesting too - we're waiting !.
Are you runing ceramic or carbon or metal film output emitter resistors - IME I don't like ceramics.
I reckon separated supplies for input and power stages ought to pay dividends - even a voltage doubler stage and then regulated to run the input stage.
IMO - this is a much better use for the Quad chassis.
More good luck, Eric.
Hi Robert,
That 'Rowland model 10 amp' is pretty well regarded isn't it ?
The bridge comparison test sounds interesting too - we're waiting !.
Are you runing ceramic or carbon or metal film output emitter resistors - IME I don't like ceramics.
I reckon separated supplies for input and power stages ought to pay dividends - even a voltage doubler stage and then regulated to run the input stage.
IMO - this is a much better use for the Quad chassis.
More good luck, Eric.
Rowland uses ICs??? and the LM3886??????
i would have thought something that high-end and expensive would have used a full discrete design. i can't believe they use the LM3886. what a cop-out. i mean i have nothing against the LM3886 and i don't doubt you if you say the Model 10 sounds great, but with a name like Jeff Rowland i would have at least expected a more original design. oh well, if Mark Levinson can get away with rebadging a $300 chinese amp and selling it for $2k with the Red Rose name...
as for emitter resistors, why not a good wirewound like Mills? i've used them and find they sound very good, crisp yet natural... better than metal film in some respects even. the 5 watt version is not too large either.
i would have thought something that high-end and expensive would have used a full discrete design. i can't believe they use the LM3886. what a cop-out. i mean i have nothing against the LM3886 and i don't doubt you if you say the Model 10 sounds great, but with a name like Jeff Rowland i would have at least expected a more original design. oh well, if Mark Levinson can get away with rebadging a $300 chinese amp and selling it for $2k with the Red Rose name...
as for emitter resistors, why not a good wirewound like Mills? i've used them and find they sound very good, crisp yet natural... better than metal film in some respects even. the 5 watt version is not too large either.
STK4050V
Robert,
I'm building an amplifier with the STK4050V module to use as a subwoofer amp. The specs only shows data for +/-66 V supply and 8 ohm loads. My supply is +/-70 V and my subs are 4 ohm so I'm wondering if I'll smoke the modules.
I saw in your post that you tried 2 ohm loads but you had a much lower supply voltage. Do you have any thoughts about what will happen with higher supply voltages?
I'm using the typ. app. schematic.
Regards, Marcus
Robert,
I'm building an amplifier with the STK4050V module to use as a subwoofer amp. The specs only shows data for +/-66 V supply and 8 ohm loads. My supply is +/-70 V and my subs are 4 ohm so I'm wondering if I'll smoke the modules.
I saw in your post that you tried 2 ohm loads but you had a much lower supply voltage. Do you have any thoughts about what will happen with higher supply voltages?
I'm using the typ. app. schematic.
Regards, Marcus
Marcus,
I just looked at the 4050V spec sheet. They quote a max of +/- 95V, so you shouldn't have any problems with it. But keep in mind at +/- 66V they're also referencing an 8 ohm load. At 4 ohms you might start to become a little limited in current delivery, but it should work.
On another matter, I tried 'voicing' the STK4040XI module. By that I varied the input cap to ground. With the 619/15k input resistors, I found 1000 pF rolled off around 120 kHz; 2000 pF at 60 kHz (-3 dB). With listening, 1000 pF + 560 pF (1560 pF) seemed best to my ears (72 kHz -3 dB point). With 2000 pF it was as smooth as the Rowland model 10. I believe this this roll off sounds best because it may limit slew distortion. At 20V/microsec and full output, the amp may start slewing, if my calcs are correct at 90 kH or so. By limiting the input to 70 to 75 kHz, this may avoid any added distortions, leading to a very smooth, rounded sound. The soundstage is still outstanding too.
Regards,
Robert
I just looked at the 4050V spec sheet. They quote a max of +/- 95V, so you shouldn't have any problems with it. But keep in mind at +/- 66V they're also referencing an 8 ohm load. At 4 ohms you might start to become a little limited in current delivery, but it should work.
On another matter, I tried 'voicing' the STK4040XI module. By that I varied the input cap to ground. With the 619/15k input resistors, I found 1000 pF rolled off around 120 kHz; 2000 pF at 60 kHz (-3 dB). With listening, 1000 pF + 560 pF (1560 pF) seemed best to my ears (72 kHz -3 dB point). With 2000 pF it was as smooth as the Rowland model 10. I believe this this roll off sounds best because it may limit slew distortion. At 20V/microsec and full output, the amp may start slewing, if my calcs are correct at 90 kH or so. By limiting the input to 70 to 75 kHz, this may avoid any added distortions, leading to a very smooth, rounded sound. The soundstage is still outstanding too.
Regards,
Robert
I'm thinking of raising the value of the emitter resistors to make life a bit easier for the output transistors (+/-70 V supply + 4 ohm load = lots of current ).
The typ. app. suggests 0.22 ohm so what happens if I put 1.0 ohm resistors in their places? Local feedback would increase, but can that cause any problems?
Will this help protect the output devices at all?
All ideas and explanations are welcome.
Regards, Marcus
).The typ. app. suggests 0.22 ohm so what happens if I put 1.0 ohm resistors in their places? Local feedback would increase, but can that cause any problems?
Will this help protect the output devices at all?
All ideas and explanations are welcome.
Regards, Marcus
Any major supplier of generic spare parts for Jap consumer gear should have them.
In Australia we pay around AUS$25-$28 from http://www.wescomponents.com/
Eric.
In Australia we pay around AUS$25-$28 from http://www.wescomponents.com/
Eric.
Dorkus:
After listening to some of the past Rowlands, marveling at the construction and lusting for one ( model 1 and model 8, for example), I was QUITE suprised (and dissappointed) also when I found out his new trend was toward small amps based on the LM3886. I figure maybe he is still making the big amps and did this to push up the value of them I can't think of any other reason except maybe to reduce product development cost so as to increase profit margin.
I do have to admit that I haven't listened to any of the newer small amps. Actually, I haven't found any dealers near me (Chicago area) that have them available for audition.
What's next, the new Rowland preamp using paralleled 5534 op amps?
mlloyd1
After listening to some of the past Rowlands, marveling at the construction and lusting for one ( model 1 and model 8, for example), I was QUITE suprised (and dissappointed) also when I found out his new trend was toward small amps based on the LM3886. I figure maybe he is still making the big amps and did this to push up the value of them
I can't think of any other reason except maybe to reduce product development cost so as to increase profit margin.I do have to admit that I haven't listened to any of the newer small amps. Actually, I haven't found any dealers near me (Chicago area) that have them available for audition.
What's next, the new Rowland preamp using paralleled 5534 op amps?
mlloyd1
dorkus said:
Some STK sources: http://www.bdent.com/; http://www.matelectronics.com/; http://www.pacificsemi.com/; http://www.onsemi.com/)
I made another measurement of the STK4048XI; dampening factor at 1kHz measured about 300. The formula I used was:
d.f. = (Vout, no load)/[(Vout, no load)-(Vout, 8ohm load)]
maybe I did it wrong?
regards, Robert
I made another measurement of the STK4048XI; dampening factor at 1kHz measured about 300. The formula I used was:
d.f. = (Vout, no load)/[(Vout, no load)-(Vout, 8ohm load)]
maybe I did it wrong?
regards, Robert
Discontinued?
First at all, txs. for the data about places to buy. Those chips appear to be just perfect for the 5.1 monitor systems I want to build. Mostly, after the invaluable data of you guys.
rljones, you did an amazing job with the oscilation issue!
But now we discover they are discontinued?
Back to the 3886?
Ouch!
Ric
First at all, txs. for the data about places to buy. Those chips appear to be just perfect for the 5.1 monitor systems I want to build. Mostly, after the invaluable data of you guys.
rljones, you did an amazing job with the oscilation issue!
But now we discover they are discontinued?
Back to the 3886?
Ouch!
Ric
I'm not so sure whether a part is in production is so important.
I bought a few pieces each of the 4044xi and 4048xi. I used one pair of the 4048s in the amp repair I described in this thread. I'll maybe build up one or two more amps and keep a couple pieces for future work. I truly don't worry about long-term availability.
As long as you don't do something outrageous with an amp, they'll last for many years. Now if you're like my friend who plugged the output of a subwoofer into my amp--or some such story--and fried both channels, you'll more likely find that the caps will go bad before the active devices. Unless one is planning on production, where you're considering the lifetime of the product and future repair-warranty issues, inventory shouldn't be a major deterrent. (Look how many people build tube amps with tubes that haven't been made since the 1920s! I guarantee that the STK module in a properly designed amp will outlast any power tube by many years; one is lucky to get 5000 hours out of just about any tube.)
In the event that you truly destroy all your stock, you'll probably, if you are still DIY'ing, most likely you'll come up some other design. (In this Quad 405ii I described here, I not only gutted it, but tried to use the STK before I solved the problems, went on to build-in an TDA7294, which worked fine until I loaned it to my friend, and now that same amp is happily playing away with STK modules. Who knows what will be inside the chassis in 5 or 10 years?
Anyhow, I wouldn't be too concerned about building up an amp with only a few parts. How many of you actually delve into how many parts are available world-wide? Don't you simply order what you need and no more? As long as you have a good power supply, you can subsitute almost any 'amp module' you care to come up with as long as it can tolerate the voltages of your power supply. Unfortunately, most op-amp type ICs only take up to about +/-35VDC, so you're limited if you've running greater voltages. The TDAs are are better, handling in the TDA7293V version up to +/-50VDC. The STKs are rather unique with some types working up to +/-80VDC.
I have one other source to add for STK modules: http://www.techsonicusa.com Their phone is 415-412-5245 (Foster City, California). There was also a place in Glendale or Pasadena, CA as I recall, dealing with Japanese parts, but I can't seem to locate their URL.
I'd better close: I have to go count my gold, I mean STK modules...
Regards, Robert
I bought a few pieces each of the 4044xi and 4048xi. I used one pair of the 4048s in the amp repair I described in this thread. I'll maybe build up one or two more amps and keep a couple pieces for future work. I truly don't worry about long-term availability.
As long as you don't do something outrageous with an amp, they'll last for many years. Now if you're like my friend who plugged the output of a subwoofer into my amp--or some such story--and fried both channels, you'll more likely find that the caps will go bad before the active devices. Unless one is planning on production, where you're considering the lifetime of the product and future repair-warranty issues, inventory shouldn't be a major deterrent. (Look how many people build tube amps with tubes that haven't been made since the 1920s! I guarantee that the STK module in a properly designed amp will outlast any power tube by many years; one is lucky to get 5000 hours out of just about any tube.)
In the event that you truly destroy all your stock, you'll probably, if you are still DIY'ing, most likely you'll come up some other design. (In this Quad 405ii I described here, I not only gutted it, but tried to use the STK before I solved the problems, went on to build-in an TDA7294, which worked fine until I loaned it to my friend, and now that same amp is happily playing away with STK modules. Who knows what will be inside the chassis in 5 or 10 years?
Anyhow, I wouldn't be too concerned about building up an amp with only a few parts. How many of you actually delve into how many parts are available world-wide? Don't you simply order what you need and no more? As long as you have a good power supply, you can subsitute almost any 'amp module' you care to come up with as long as it can tolerate the voltages of your power supply. Unfortunately, most op-amp type ICs only take up to about +/-35VDC, so you're limited if you've running greater voltages. The TDAs are are better, handling in the TDA7293V version up to +/-50VDC. The STKs are rather unique with some types working up to +/-80VDC.
I have one other source to add for STK modules: http://www.techsonicusa.com Their phone is 415-412-5245 (Foster City, California). There was also a place in Glendale or Pasadena, CA as I recall, dealing with Japanese parts, but I can't seem to locate their URL.
I'd better close: I have to go count my gold, I mean STK modules...
Regards, Robert
one other resource is: http://audiolab.home.mindspring.com/ (any 'xi' version is fully complementary and should work as described in this thread), eg, stk4044xi.
on the audiolab web site, they have an stk data sheet section: http://www.datadart.com/commercesql/index.cgi?page=STK Datasheets.html&cart_id=5901407.90681
here you can look up the '*.xi' ics.
regards, robert
on the audiolab web site, they have an stk data sheet section: http://www.datadart.com/commercesql/index.cgi?page=STK Datasheets.html&cart_id=5901407.90681
here you can look up the '*.xi' ics.
regards, robert
Narrowing the choices is the game's name
Nice links, rljones,
One thing I've noticed is the wild difference in price between vendors. Pacific Semiconductor sells the 4048Xi for $11.39 but Techsonic sells it for $21.
With this information I'm narrowing the choices for my 5.1 active monitors to’ one 4044XI for the highs and one 4048 for lows. That would be perfect.
I still have to decide on the design of the loudspeakers, but that's another thread in the loudspeakers sub-forum.
Anyway, txs a lot for the quality information.
Cheers
Ric
Nice links, rljones,
One thing I've noticed is the wild difference in price between vendors. Pacific Semiconductor sells the 4048Xi for $11.39 but Techsonic sells it for $21.
With this information I'm narrowing the choices for my 5.1 active monitors to’ one 4044XI for the highs and one 4048 for lows. That would be perfect.
I still have to decide on the design of the loudspeakers, but that's another thread in the loudspeakers sub-forum.
Anyway, txs a lot for the quality information.
Cheers
Ric
I made up a bridged pair of STK4048XI. To test it out, I built up new boards (which I'll upload as soon as I figure out how to transfer them from my PC to my Mac in a legible format for uploading; I'm too ignorant of PCs to know how to do it directly, and help here?).
My test power supply uses 25VAC secondaries, generating around +/- 33VDC. Each module produced around 65W to 70W into 8 ohms and around 110W/4 ohms in standard configuration.
With this same power supply, in bridged mode, using a Jensen JT-11P-1 transformer as a phase splitter (see their web site for details), the amp on the test bench produced 190W/8 ohms (39V RMS) and 270W/4 ohms (33V RMS). With a more typical +/-49VDC supply, I would expect around 300W/8 ohms and 450W/4 ohms. I don't have a balanced distortion analyzer, so I didn't make any measures aside from power readings and looking at the waveforms on the scope. I don't believe that I would like to run them at +/-55VDC in bridged mode as the power dissipation goes up dramatically: a bridged 8 ohm load is like a 4 ohm load to an amp in terms of current demand, and a bridged 4 ohm load is like a 2 ohm load (low impedance loads are a lotta work for a bridged amp).
I did learn one other thing. The boards initially oscillated and the 200pF cap in the feedback area didn't do squat. I eliminated the problem by sequentially testing all caps in the app note in various purmutations. Refering to the app note for the STK4048XI, I changed C6, C7 and C9 each from 100pF to 200pF and this solved the problem and required no 200pF on the feedback scheme previously described in this thread. One other capacitor was also important in the bridged mode to prevent oscillation: C1. On the app note C1 is 470pF in parallel with 56kohms (R2). I needed to up this to 1,560pF in parallel with 20kOhms (and 619 ohms for R1 instead of 1k). Only using 1,000pF for C1, resulted in oscillation in bridged mode even with the 200pF changes to C6, C7, and C9. Therefore, if you see oscillation, which will probably vary with layout, play with these 4 capacitors.
Regards, Robert
My test power supply uses 25VAC secondaries, generating around +/- 33VDC. Each module produced around 65W to 70W into 8 ohms and around 110W/4 ohms in standard configuration.
With this same power supply, in bridged mode, using a Jensen JT-11P-1 transformer as a phase splitter (see their web site for details), the amp on the test bench produced 190W/8 ohms (39V RMS) and 270W/4 ohms (33V RMS). With a more typical +/-49VDC supply, I would expect around 300W/8 ohms and 450W/4 ohms. I don't have a balanced distortion analyzer, so I didn't make any measures aside from power readings and looking at the waveforms on the scope. I don't believe that I would like to run them at +/-55VDC in bridged mode as the power dissipation goes up dramatically: a bridged 8 ohm load is like a 4 ohm load to an amp in terms of current demand, and a bridged 4 ohm load is like a 2 ohm load (low impedance loads are a lotta work for a bridged amp).
I did learn one other thing. The boards initially oscillated and the 200pF cap in the feedback area didn't do squat. I eliminated the problem by sequentially testing all caps in the app note in various purmutations. Refering to the app note for the STK4048XI, I changed C6, C7 and C9 each from 100pF to 200pF and this solved the problem and required no 200pF on the feedback scheme previously described in this thread. One other capacitor was also important in the bridged mode to prevent oscillation: C1. On the app note C1 is 470pF in parallel with 56kohms (R2). I needed to up this to 1,560pF in parallel with 20kOhms (and 619 ohms for R1 instead of 1k). Only using 1,000pF for C1, resulted in oscillation in bridged mode even with the 200pF changes to C6, C7, and C9. Therefore, if you see oscillation, which will probably vary with layout, play with these 4 capacitors.
Regards, Robert