Ceci est un forum anglais. Si vous souhaitez utiliser une autre langue, le poster en anglais ainsi.
This is an English forum. If you wish to use another language, post it in English as well.
yoyo44 via Google Translate:
bjr I want a picture of the back of an amp brilliantly 1 please
because mine was hack and bakelite abyss
thank you to you
good day
Linear Audio Vol. 9 [published 1 April 2015] has an interesting article which is entitled A Low Count Audiophile Power Amplifier by Vanderkooy et al. The design of this amp may fit the stated objective.
N.B My apology for putting this note here instead of the thread called Composite Amplifier Design. Please ignore. Thank you
N.B My apology for putting this note here instead of the thread called Composite Amplifier Design. Please ignore. Thank you
Last edited:
If I was teaching audio electronics I might use the PE Texan as a way to teach as much as possible in a short time. Not least star grounding and how an amplifer output stage might have voltage gain. Then biasing and op amp chioce. The final thing would be to ask the students if they could get it to sound good whilst saying why it might be difficult. Most of all how to measure it. An 8 watt Texan might be very OK ( gain of 1 ). I would also say how the PE Orion was the step forward and look that we might try to build it with and without a long tail pair input ( perhaps without would have the sonic advantage ). Then say how the op amp can be used as a preamp device and it's origins in analogue computing ( modern multi meter ). I do like the positioning of the op amp and feedback type ( not the filtering ).
A Paul Kemble web page - Practical Electronics Gemini and Orion amplifiers.
It is a wonder Douglas Self didn't try a little voltage gain in his usual designs if only to say why not ( maybe he did ). I used a gain of 3 and 2SA1962/2SC5242 and LF351/N as what was in the useful box. It all looked very OK on the analyser even into a very reactive load. I don't see the need for higher output stage gain than I used. A gain switch looks an idea ( in my case 1 or 3 ). Hope this gets someones interest and isn't too late. Mine might get used to drive my 15 inch unit that rolls off at 250 Hz.
A Paul Kemble web page - Practical Electronics Gemini and Orion amplifiers.
It is a wonder Douglas Self didn't try a little voltage gain in his usual designs if only to say why not ( maybe he did ). I used a gain of 3 and 2SA1962/2SC5242 and LF351/N as what was in the useful box. It all looked very OK on the analyser even into a very reactive load. I don't see the need for higher output stage gain than I used. A gain switch looks an idea ( in my case 1 or 3 ). Hope this gets someones interest and isn't too late. Mine might get used to drive my 15 inch unit that rolls off at 250 Hz.
Hi Nigel, I think you'll find DS does discuss CFP designs with gain, generally. I found a page plus schematics in the 5th ed. of his APADH, p 149. He's not particularly averse but concludes that if CFP with gain was a design option, his advice is that you probably don't want to go this way.
Last edited:
Hi Ian. Seeing as a compound pair even with gain will be highly linear there might be a sweet spot where transfering the gain duty to the output stage seems not a bad idea. I speculate a gain of 3 would be ideal.
Using one of Self's arguements. We only have loop gain to conquer crossover distortion thus we should attempt the highest loop gain which we then control by Cdom ( sorry if that's not exactly his words ) . The compound pair as I think some call the Texan/Brio looks to be the one thing Self should really love. If you stop to think having extra voltage gain at the output is unlikely to make the amplifier more unstable. My instinct is to think the opposite. It seems a shame the early origines and use if the 741 made the Texan a bit of a joke. My tests of the Texan generic concept saw a most unusual trait. It very nearly is a rail to rail design with quite reasonable clipping considering that. As far as I could see the distortion is low right up to that point. I dare say many will hate it for that, maybe even me. I could imagine a low distortion valve preamp would suit it. That is to balance off it's distortion signature. I could imagine issues of slewing might be helped by relaxing the VAS a little. The PE Orion hints at this.
Using one of Self's arguements. We only have loop gain to conquer crossover distortion thus we should attempt the highest loop gain which we then control by Cdom ( sorry if that's not exactly his words ) . The compound pair as I think some call the Texan/Brio looks to be the one thing Self should really love. If you stop to think having extra voltage gain at the output is unlikely to make the amplifier more unstable. My instinct is to think the opposite. It seems a shame the early origines and use if the 741 made the Texan a bit of a joke. My tests of the Texan generic concept saw a most unusual trait. It very nearly is a rail to rail design with quite reasonable clipping considering that. As far as I could see the distortion is low right up to that point. I dare say many will hate it for that, maybe even me. I could imagine a low distortion valve preamp would suit it. That is to balance off it's distortion signature. I could imagine issues of slewing might be helped by relaxing the VAS a little. The PE Orion hints at this.
To get stability I used the crudest way to get it. A RC filter directly from op amp out to bias chain in. My bias was the crudest also which in simplest form is 2 x 1N4148 with the op amp fed to the middle adjusted by the chain current ( 2 x 18K when Brio ). I was only interested in stability at that point so not saying it's the ideal way.
If I had a Brio of this era I would try reducing output gain to 3.5 or whatever, the op amp will take up the slack. I would try to fit a larger transformer if the case allows . Then fit op amp sockets to see where that goes, I used LF351N as it was what was around. My version has inverting gain so as to sum the feedback at the input, I have tried the non inverting and found it noticably less stable. Instead of taking the +ve input directly to ground I use a resistor to balance the input currents ( for no obvious gain, DC offset no doubt ).
If anyone ever tried to make a better Texan it would be a good thread as it really would teach the basics. I have a hunch it is a good design in search of a little thought. Mine was built on a piece of Veroboard of 1 x 2 inch with the transistors deadbug. One thing I really noticed. The supply decoupling is very important, more so than most. I could imagine a double feedback route to the op amp might work, one to itself as if just an op amp ( 1/3? Mine has this for a completely different reason ). Seeing as a gain of 1 compound pair has almost zero distortion compared with the usual Darlington or FET a little output gain and two loops should not killl it's hi fi ability, a little damping factor change should be OK. What I would try to avoid is the various filtering options the original Texan used, using my op amp out filter seems to be better. My favoured transistors are BD139/140 ( 16 ) and 2SA1962/2SC5242, gain is the main reason. I got the idea of the op amp output filter from a Rotel amp that I like the sound of. It's the no-no example in Self where the VAS is loaded down, subtly different here as the op amp should have the spare current to drive it. The more I think about it I am not so sure Cdom as it is called is my favourite, JLH was saying the same when his 1980 bipolar plus audio FET compound pair ( was replacing a compound triple ).
The Texan/Brio distortion type is unfortunately the " tipical transistor sound circa 1970 ". However I am not convinvinced that's the problem. I suspect the Texan was always unstable on real music. The cure might have degraded the last digit of THD, however it would sound better. I ran a very reactive load on mine, that's the test.
If I had a Brio of this era I would try reducing output gain to 3.5 or whatever, the op amp will take up the slack. I would try to fit a larger transformer if the case allows . Then fit op amp sockets to see where that goes, I used LF351N as it was what was around. My version has inverting gain so as to sum the feedback at the input, I have tried the non inverting and found it noticably less stable. Instead of taking the +ve input directly to ground I use a resistor to balance the input currents ( for no obvious gain, DC offset no doubt ).
If anyone ever tried to make a better Texan it would be a good thread as it really would teach the basics. I have a hunch it is a good design in search of a little thought. Mine was built on a piece of Veroboard of 1 x 2 inch with the transistors deadbug. One thing I really noticed. The supply decoupling is very important, more so than most. I could imagine a double feedback route to the op amp might work, one to itself as if just an op amp ( 1/3? Mine has this for a completely different reason ). Seeing as a gain of 1 compound pair has almost zero distortion compared with the usual Darlington or FET a little output gain and two loops should not killl it's hi fi ability, a little damping factor change should be OK. What I would try to avoid is the various filtering options the original Texan used, using my op amp out filter seems to be better. My favoured transistors are BD139/140 ( 16 ) and 2SA1962/2SC5242, gain is the main reason. I got the idea of the op amp output filter from a Rotel amp that I like the sound of. It's the no-no example in Self where the VAS is loaded down, subtly different here as the op amp should have the spare current to drive it. The more I think about it I am not so sure Cdom as it is called is my favourite, JLH was saying the same when his 1980 bipolar plus audio FET compound pair ( was replacing a compound triple ).
The Texan/Brio distortion type is unfortunately the " tipical transistor sound circa 1970 ". However I am not convinvinced that's the problem. I suspect the Texan was always unstable on real music. The cure might have degraded the last digit of THD, however it would sound better. I ran a very reactive load on mine, that's the test.
I think some people did quite like the original Brio and clamshell versions. Here's a link to is Stereophile's review from 1998. Note the comment that the Texan type design was chosen by audition of 8 different designs presented by Terry Bateman, Rega's design engineer. Also, there is a hint about the stability there in the specification of "low tech" transistors.
https://www.stereophile.com/integratedamps/998rega/index.html
https://www.stereophile.com/integratedamps/998rega/index.html
Hi Ian, The story I remember was that Terry came in to rescue an amplifier that had been designed by one of the then guru's. Terry said at the time Roy his boss was pulling his hair out because no two amplifiers sounded the same when that design. Terry simplified the circuit and built test gear to ensure each would be the same. As I understand it Roy asked Terry as his next project if he could produce the simplest possible design and the Texan with improvements was it. It was an entry level product, The Brio.
Roy was the one who was involved in the fine tuning and most likely did have 8 plus versions given to him ( I think Terry said he got so fed up with the questions that the 8 versions were to find the ideal variation ). I can believe this as Roy is known to have listened to nylon verses steel screws on the Rega motor mounts for some weeks. Nylon was prefered as the customers could not over tighten them, he wasn't sure if it sounded as good! I saw the test gear he built.
Roy was the one who was involved in the fine tuning and most likely did have 8 plus versions given to him ( I think Terry said he got so fed up with the questions that the 8 versions were to find the ideal variation ). I can believe this as Roy is known to have listened to nylon verses steel screws on the Rega motor mounts for some weeks. Nylon was prefered as the customers could not over tighten them, he wasn't sure if it sounded as good! I saw the test gear he built.
I guess industry scuttlebutt is more credible if you have encountered the people involved or seen more of their work. What comes out of Rega's published comments on the Brio design I think, is the likely problems of using high Ft transistors like the 2SA1962/2SC5242 pair where there is considerable gain.
The CFP configuration is used with such transistors in Rod Elliott's simple P3a and a few other designs without many problems, provided a bit of shunt capacitance is added across the negative side driver's Collector/Base junction.
If you are bent on using fast output transistors in the amplifier, at least compare stability with an TIP3055/2955 pair in the same prototype. The 2SC5242/A1962 pair look to have similar specs to the ubiquitous 2SC5200/A1943 pair but in a TO3P package, like 2SC5200N/A1943N. I'm fairly sure that the extra Ft on offer is the likely cause of instability but let's see just how much.
The CFP configuration is used with such transistors in Rod Elliott's simple P3a and a few other designs without many problems, provided a bit of shunt capacitance is added across the negative side driver's Collector/Base junction.
If you are bent on using fast output transistors in the amplifier, at least compare stability with an TIP3055/2955 pair in the same prototype. The 2SC5242/A1962 pair look to have similar specs to the ubiquitous 2SC5200/A1943 pair but in a TO3P package, like 2SC5200N/A1943N. I'm fairly sure that the extra Ft on offer is the likely cause of instability but let's see just how much.
Last edited:
Ian. I think my simple RC filter from op amp out to bias chain in is reliable. The current gain isn't a problem, nor really ft. If capaitance is added to the output devices one gets the current gain and a controlable Cob or whatever. I have also used BD139/140 TIP35/36 with good results.
To be honest this circuit is unstable even when all the original design features are fitted and slower devices. The op amp output RC filter of 100R + 33 nF is a good place to start. 10 nF with TIP35/36 seems about right. You must use a scope and a reactive load, I'm told simulations will do the same ( really ? ). I will look at Rod's P3, I think I remember it. I see no harm in adding a cog-npo capacitor to each output or driver.
To be honest this circuit is unstable even when all the original design features are fitted and slower devices. The op amp output RC filter of 100R + 33 nF is a good place to start. 10 nF with TIP35/36 seems about right. You must use a scope and a reactive load, I'm told simulations will do the same ( really ? ). I will look at Rod's P3, I think I remember it. I see no harm in adding a cog-npo capacitor to each output or driver.
One idea that could work is a class A mode with unity gain of 8 watts and AB of whatever you like. One could use +/- 15V ( class A ) or +/- 30V or even +/- 45 V. The latter could be 3 x +/-15 V. The class A could have a thermal switch to throw it into AB should it get too hot ( a resistor is enough ). Octal plugs might do the connections. This also gives the class A plenty of uF. Has anyone seen a 100 watt 8R version?
One nice side to the class A and this amp is it is likely to swing near the rail. If it was > 2 watts A and the rest AB mostly Gm doubling is then a non issue. Speakers of >87 dB/W seldom need more than 2 watts for detailed music . 100 watts is nice for party mode.
Rega said that the Brio was picked out on blind test. It's specs would have put most people off. I had a Brio. It wasn't really able to party like NAD's Rotels or even Arcams. The Creek was cheaper and well liked despite similar traits, price is king. I always felt the Brio was penny pinched as the reason why, transformer perhaps. If the Brio had a wonderful 5 watts I would have loved it, the build was exactly to my taste. The party duties to the Rotel's. Yamaha's on paper were best of type. In real life the NAD sounded louder although claiming less than half the power. The Brio should give high output current if it's there to give. Like the Citroen 2CV the Brio is such a logical piece of engineering. It is worth discussion.
One idea that might have value is a LM1875 or whatever doing the op amp bit and underbiased outputs with some feedforward from the LM1875. These designs seldom are what the Quad 405 suggests ( Nor the 405 I think ). All the same, it might work. I suspect the outputs biased at 1 V would be about right.
To repeat myself. You should find inverting input gives best stability. If anything like me it will be quite noticable. If so it must sound better.
One idea I must try is BD139/140 with IRF640/9640 which I have tons of. It should be trouble from the word go.
BTW. This is great repair advice as the use of slower output devices is the best advice. Problem is even the slower devices of now are faster than then. I am told TIP2955/3055 often is whatever device is closest of silicon the company has. TIP35/36 looks very similar.
One nice side to the class A and this amp is it is likely to swing near the rail. If it was > 2 watts A and the rest AB mostly Gm doubling is then a non issue. Speakers of >87 dB/W seldom need more than 2 watts for detailed music . 100 watts is nice for party mode.
Rega said that the Brio was picked out on blind test. It's specs would have put most people off. I had a Brio. It wasn't really able to party like NAD's Rotels or even Arcams. The Creek was cheaper and well liked despite similar traits, price is king. I always felt the Brio was penny pinched as the reason why, transformer perhaps. If the Brio had a wonderful 5 watts I would have loved it, the build was exactly to my taste. The party duties to the Rotel's. Yamaha's on paper were best of type. In real life the NAD sounded louder although claiming less than half the power. The Brio should give high output current if it's there to give. Like the Citroen 2CV the Brio is such a logical piece of engineering. It is worth discussion.
One idea that might have value is a LM1875 or whatever doing the op amp bit and underbiased outputs with some feedforward from the LM1875. These designs seldom are what the Quad 405 suggests ( Nor the 405 I think ). All the same, it might work. I suspect the outputs biased at 1 V would be about right.
To repeat myself. You should find inverting input gives best stability. If anything like me it will be quite noticable. If so it must sound better.
One idea I must try is BD139/140 with IRF640/9640 which I have tons of. It should be trouble from the word go.
BTW. This is great repair advice as the use of slower output devices is the best advice. Problem is even the slower devices of now are faster than then. I am told TIP2955/3055 often is whatever device is closest of silicon the company has. TIP35/36 looks very similar.
Several or more years ago now I bought a Rega CD player and decided to match this with a Brio amplifier.
I was able to audition the latter on approval but ended up exchanging it for a Nait 5i on the same basis.
In the end I bought this and traded the Rega CD player (at a substantial loss for the time I owned it) for the Naim CD5i-2.
The Brio was OK for the asking price and it was reliable. Both Rega pieces had plastic cases and lacked remote controls. The paintwork is easily damaged. These things worked against my staying with them.
All this came about when the glass door on the cabinet was closed when the tray of my Cambridge CD player was open and it became jammed. I thought it time for an upgrade - I managed to find a repairer and still have this player. But for my impulsiveness at the time I might have bought a Cambridge amplifier and been equally as well off.
This is the Brio amplifier. I am a little puzzled about the voltage dividers R18/R20 and R23/R21 which set output stage gain to 11 when ~2.5 would suffice for rail voltages of +/- 32V whilst ever the opamp is tied to +/-15V rails. Or have I missed something that should be obvious ?
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.