The op.amp output will swing to around +/-14 volts with a 10k load or 13 volts with 2k.
It is very interesting to play with that idea. For 100 watts it makes more sense. I found it very much swings and roundabouts playing with the gain. My speculation is this. If output gain is 10 perhaps the amplifier is more stable. My feeling is to make two feedback loops and an output gain of 3 is an upgrade. The first feedback loop on the op amp, hopefully more is won than lost. The +/-15 V is not required as is. I only have two resistors to give my output gain. The ones I see in the Brio and Texan I think are where the designers used output resistors ( 0R33 or whatever ) and then changed their minds if Texan mk1. Or am I missing something. Mine works fine when 1 x 82R and 1 x 240R. Maybe the +/-15 V was from a time when the output gain was tried. Maybe like me the designer retains the phases of a design as a memory of the journey.
If I have time I will build a 100 watt version. It should be fine. Like Terry and Roy at Rega I will listen to it with an open mind. I have a 500VA 0-25 0-25 that almost would do it as we get about 0-28 0-28 when 246 VAC at home. I also have some 20 000 uF 40V which will do for now.
Hi!
I picked up a little clamshell brio for a few euros. One channel has a dead power amplifier, unfortunately it had been worked on and someone replaced a bunch of resistors (mostly wrong values) and all the transistors (correct references but probably fakes for some of them). The outputs and the bc546 aren't a problem but I would like to find a replacement for the BC639/640 driver pair, even the on semi version from mouser is on order and they don't have the same gain/power versions. Is there a drop in replacement or even something better without modifying the rest of the circuit ?
I picked up a little clamshell brio for a few euros. One channel has a dead power amplifier, unfortunately it had been worked on and someone replaced a bunch of resistors (mostly wrong values) and all the transistors (correct references but probably fakes for some of them). The outputs and the bc546 aren't a problem but I would like to find a replacement for the BC639/640 driver pair, even the on semi version from mouser is on order and they don't have the same gain/power versions. Is there a drop in replacement or even something better without modifying the rest of the circuit ?
That was the advice I would have given. The specs of BD's are very weird, long ago >100 MHz, now anyones guess. I suspect some are MJE340/350 which are very slow. As the great Douglas Self uses 340/350 without any warnings they should be OK.
Here below are some parts that might be worth having. They seem to resemble 2SB716/2SD756 which mostly seem to be fakes now and silly money. I suspect the base and emmiter are opposite to BC639/640 with same centre collector so easy to fit. If you look these are do everything devices. The low current gain is quite typical and is often far better when in an amplifier at 8 mA or whatever. They are very much like better versions of the BC/BD devices of the original Philips specs. They look to me they would sound good as I suspect they were TV CRT drivers which often were in class A without feedback. They were linear enough to avoid colour changes in shadows, that is a blue, green or pink tinge that shifted with brightness.
Watch the Brio. It can go unstable and needs setting up on a scope. Not a defect really. PW Texan and PE Rondo are the related designs. The output stage has gain. In my view too much. Having said that I have driven a large induction motor with one once set up with no problems at all.
I suspect the Brio would sound better with a same voltage larger transformer.
I asked Rega's engineer why what looked to be emitter resistors were so large in value. He said because the values became cheaper at that value and was a small compromise. That's not exactly true. It is the gain setting.
A Paul Kemble web page - PW Texan amplifier and the PE Rondo.
KSA916YTA | ON Semi KSA916YTA PNP Transistor, 800 mA, 120 V, 3-Pin TO-92 | ON Semiconductor
KSC2383YTA | ON Semi KSC2383YTA NPN Transistor, 1 A, 160 V, 3-Pin TO-92 | ON Semiconductor
Here below are some parts that might be worth having. They seem to resemble 2SB716/2SD756 which mostly seem to be fakes now and silly money. I suspect the base and emmiter are opposite to BC639/640 with same centre collector so easy to fit. If you look these are do everything devices. The low current gain is quite typical and is often far better when in an amplifier at 8 mA or whatever. They are very much like better versions of the BC/BD devices of the original Philips specs. They look to me they would sound good as I suspect they were TV CRT drivers which often were in class A without feedback. They were linear enough to avoid colour changes in shadows, that is a blue, green or pink tinge that shifted with brightness.
Watch the Brio. It can go unstable and needs setting up on a scope. Not a defect really. PW Texan and PE Rondo are the related designs. The output stage has gain. In my view too much. Having said that I have driven a large induction motor with one once set up with no problems at all.
I suspect the Brio would sound better with a same voltage larger transformer.
I asked Rega's engineer why what looked to be emitter resistors were so large in value. He said because the values became cheaper at that value and was a small compromise. That's not exactly true. It is the gain setting.
A Paul Kemble web page - PW Texan amplifier and the PE Rondo.
KSA916YTA | ON Semi KSA916YTA PNP Transistor, 800 mA, 120 V, 3-Pin TO-92 | ON Semiconductor
KSC2383YTA | ON Semi KSC2383YTA NPN Transistor, 1 A, 160 V, 3-Pin TO-92 | ON Semiconductor
That's a lot more information than I was hoping for but in the end, should I just go for the BD139/140 (indeed specs are almost exactly the same) or try either the 2sb716/2sd756 and KSA916/KSC2383 combinations and see what works best ? My goal is to keep it as close as possible to original.
Does anyone know if the recent version of the Brio still use these as drivers or if they replaced them with something easier to find. From what I can see the schematic didn't change from the original design.
Does anyone know if the recent version of the Brio still use these as drivers or if they replaced them with something easier to find. From what I can see the schematic didn't change from the original design.
I am sorry to say your guess is as good as mine what works best. No one really can say if the devices we buy today are the same as the past ansd usually in a bad way. 2SD756/b716 , just forget them as they cost a bomb and will often ( always ) be fakes. They resemble 2N5401/5551 in big TO92 package. If you don't mind the difficult collector location 5401/5551 can work very well.
The thing you have to understand is the combination of an op amp to control the power amp and an output stage with gain is a disaster waiting to happen. If you see plenty of smoke on start up that's what I mean. With a little care it need not happen. In fact the amplifier is very stable once the stability point is found. If I had never seen a Brio or Texan amp I would have said " that will not work ". Fixing amplifiers is not like fixing engines, they can be bad enough. All amplifiers I know of make excellent oscillators and given the chance will do that.
Rega often are very helpful. Ask them.
The thing you have to understand is the combination of an op amp to control the power amp and an output stage with gain is a disaster waiting to happen. If you see plenty of smoke on start up that's what I mean. With a little care it need not happen. In fact the amplifier is very stable once the stability point is found. If I had never seen a Brio or Texan amp I would have said " that will not work ". Fixing amplifiers is not like fixing engines, they can be bad enough. All amplifiers I know of make excellent oscillators and given the chance will do that.
Rega often are very helpful. Ask them.
Here is a version of the Brio. Notice it doesn't have the frequecy compensation feedback loop of the Texan. This makes it more likely to protest. The 560/56 gain setting is excessive in my opinion. However it is an exercise is negative feedback I guess. I am a bit surprised the amp can work with this simplicty. I repeat my warnings not to get too careless with choices. TL072 seems a very wise choice. The red bits are the person who posted it choice. I have built this type of amp with an output gain of three and liked it, even two worked.
Looking again I guess C6 R25 are looking to do the same.
Last edited:
I found some BD139-10 and BD140-10 from on-semi on mouser. the higher hfe version is once again on order (shipping date january 2019, I'll pass). I always power up and do a first test with a dim bulb, I'll scope the output to look for oscillation. Some suggested to change the op amp for a ne5532, but you would stick with the original TL072 ?
I don't think its a good idea to use higher hFE range parts than specified in the Brio. Having watched a couple of boards burn as I was testing them, I can justify stability warnings. I would only use standard or -10 drivers, unless you like the chance of a heat and smoke finale with your music 
We should also be careful how we interpret hFE measurements. The cheap Ebay parts testers seldom give results that relate to manufacturer specs, as their test current is about 2mA max. Results can therefore be anywhere, depending on the Hfe V Ic characteristic curve of the transistor. The table shows differences for BD139 at more realistic currents. BD140 is quite similar. As jaycee suggests, anything apart from heat dissipation that applies to BC639/640, will also apply to BD139/140.Philips - datasheet pdf
Hi Depaj, I'm sure you know that you can buy ST Micro BD139/140 products from Element 14, RS components, Profusion, Reichelt and other distributors in Europe. Mouser seem to deal with a lot of suppliers but now they hold very few stock lines of useful through-hole parts. It's annoying and FWIW, they have lost my business.
We should also be careful how we interpret hFE measurements. The cheap Ebay parts testers seldom give results that relate to manufacturer specs, as their test current is about 2mA max. Results can therefore be anywhere, depending on the Hfe V Ic characteristic curve of the transistor. The table shows differences for BD139 at more realistic currents. BD140 is quite similar. As jaycee suggests, anything apart from heat dissipation that applies to BC639/640, will also apply to BD139/140.Philips - datasheet pdf
Hi Depaj, I'm sure you know that you can buy ST Micro BD139/140 products from Element 14, RS components, Profusion, Reichelt and other distributors in Europe. Mouser seem to deal with a lot of suppliers but now they hold very few stock lines of useful through-hole parts. It's annoying and FWIW, they have lost my business.
Attachments
Last edited:
FET op-amps work best in DC feedback circuits. Unless you like having your output slammed against one of the rails, leave it as a TL072.
If you want it tweakable/swappable, separate that function from the input amplifier circuitry.
The KS parts I listed should work out if the base emmiter direction is factored in. As far as I know centre collector with reversed b,e. Thus they are 180 degrees rotated.
If you note all they were was good versions of the BC639/640 in style.
TL072 seems to work better if used in a power amp. I would sepeculate the JFET input and resistor output current limiting helps. NE5532 might be trouble.
CDIL BC639 Transistor NPN 80V 1A TO92 | Rapid Online
CDIL BC640 TO92 80V PNP Low Frequency Transistor | Rapid Online
If you note all they were was good versions of the BC639/640 in style.
TL072 seems to work better if used in a power amp. I would sepeculate the JFET input and resistor output current limiting helps. NE5532 might be trouble.
CDIL BC639 Transistor NPN 80V 1A TO92 | Rapid Online
CDIL BC640 TO92 80V PNP Low Frequency Transistor | Rapid Online
From a Different Perspective
There is a consistent pattern of burned out resistors in three Brio amplifier cases where the suggested solution is to replace some transistors with parts rated for higher power.
The closed loop gain of the whole is 150k/1k plus one which is the minimum gain of a series feedback system. This is to obviate the need for any line stage.
It has been suggested that the gain of the discrete component stage of 11 times could be reduced.
Unfortunately the gain of a TL072 declines to one at 3MHz so if the entire amplification relies on this, the point at which the gain reduces to one is 3MHz divided by 150 or 20kHz.
A consequence of increasing the closed loop gain (discrete stage included) increasing sensitivity for input sources also heightens that to any disturbance on the feedback line.
Fuses are undesirable in speaker lines and one way to deal with this is to include them in the feedback loop and run a resistor in parallel to maintain dc conditions.
There is a snag in this scheme as the fuse is thus a part of the signal source so if it is caused to blow in some extreme event the effect is going to be magnified 150 times.
Further in this instance there is a 22 p lead compensation capacitor in the feedback line which lowers the impedance path to the feedback terminal at high frequencies.
I attribute the burned out resistors to instability caused by fuses blowing under conditions of drive heavy enough to blow output fuses. Increasing fuse ratings, resistor ratings, or transistor ratings is not the answer to this.
My solution is to place the 22p capacitor and place it between the inverting input of the op.amp and its' output providing a feedback path at high frequencies that will not be directly influenced by such as a blown output fuse.
There is a consistent pattern of burned out resistors in three Brio amplifier cases where the suggested solution is to replace some transistors with parts rated for higher power.
The closed loop gain of the whole is 150k/1k plus one which is the minimum gain of a series feedback system. This is to obviate the need for any line stage.
It has been suggested that the gain of the discrete component stage of 11 times could be reduced.
Unfortunately the gain of a TL072 declines to one at 3MHz so if the entire amplification relies on this, the point at which the gain reduces to one is 3MHz divided by 150 or 20kHz.
A consequence of increasing the closed loop gain (discrete stage included) increasing sensitivity for input sources also heightens that to any disturbance on the feedback line.
Fuses are undesirable in speaker lines and one way to deal with this is to include them in the feedback loop and run a resistor in parallel to maintain dc conditions.
There is a snag in this scheme as the fuse is thus a part of the signal source so if it is caused to blow in some extreme event the effect is going to be magnified 150 times.
Further in this instance there is a 22 p lead compensation capacitor in the feedback line which lowers the impedance path to the feedback terminal at high frequencies.
I attribute the burned out resistors to instability caused by fuses blowing under conditions of drive heavy enough to blow output fuses. Increasing fuse ratings, resistor ratings, or transistor ratings is not the answer to this.
My solution is to place the 22p capacitor and place it between the inverting input of the op.amp and its' output providing a feedback path at high frequencies that will not be directly influenced by such as a blown output fuse.
A lot has been going on since the last message I saw, I'll have to take some time to go through all of this. I was wondering though, didn't rega come up with a solution for these issues in their newer versions of the Brio or did they just completely modify the whole circuit ?
The TL072 doesn't seem to have taken a hit when the whole thing went up in smoke, I was just curious as to why some suggested the NE5532.
I have bought things from the first 2, thing is, as much as possible I try to buy from the same place to reduce shipping and mouser does have a lot of parts that element 14 or rs don't have. Is ST Micro superior in quality than on semi or do you suggest them because the datasheet has more info ?
You mean remove the 22pF cap at the end and place it directly on the TL072 ?You seem to have repaired a few so I assume you have tried this, does it affect the sound in any way or is this really just to increase stability ?
The TL072 doesn't seem to have taken a hit when the whole thing went up in smoke, I was just curious as to why some suggested the NE5532.
I have bought things from the first 2, thing is, as much as possible I try to buy from the same place to reduce shipping and mouser does have a lot of parts that element 14 or rs don't have. Is ST Micro superior in quality than on semi or do you suggest them because the datasheet has more info ?
You mean remove the 22pF cap at the end and place it directly on the TL072 ?You seem to have repaired a few so I assume you have tried this, does it affect the sound in any way or is this really just to increase stability ?
I would not worry over the choice of brands between first rank manufacturers. You should be able to use either and source them from element 14 or RS Components.
In this country there is no charge for freight for RS Component account holders - the downside is much of their range is sold in minimum quantities.
There is a full schematic for the Brio on Hi-Fi Engine in case you are curious about the phono stage.
I had a Brio 3 for a few months but traded it in for a Nait5i so although I have repaired some amplifiers for family and friends this has not included a Brio 3.
The TL072 has not taken a hit since this has inbuilt protection and there is a 100R resistor in series with the output that will lessen the chances of that being triggered.
The NE5532 has a BJT input and there is no blocking input capacitor so the dc output will vary according to the volume control setting - a point possibly overlooked by those who suggested the substitution.
An LF353 is an equivalent to TL072.
To my ears the former has a brighter sound and it might be a better pairing with the changed form of compensation that tends to tone down this characteristic.
To implement the change you could look for some NP0 chip capacitors, and after tinning, solder them between pins 1 and 2 and pins 6 and 7.
The changes before and after would have to be tested by simulations.
There has to be a compromise between THD which is measured on sine waves and transient distortions which might show up as a spike or ripple on square wave signals. To strike a balance a telling measure would be settling time.
My concern is more about mitigating risks of component failure during a malfunction. The closed loop gain of this unit is several times greater than normal and a fuse is part of that circuit.
As far as I know, only the first (clamshell) Brio uses this particular boosted OP-AMP design. Later models, have been described as tweaked versions but I don't think so. Brio 3 for example, has SAP darlington output transistors and an LTP input stage. The op-amps are only in the preamp, as far as I can see.http://www.audiocostruzioni.com/r_s...brio-3/image-r937x635-C-ba672373-12983387.jpg
I haven't yet seen other models like Brio 2000 and brio 2 up close, so I can't really comment further. The current Brio-R model though, has a completely different design again, said to have more in common with Elex/Elicit models.
Based on your first choice at Mouser, I assumed you had a preference for ST Micro product. I sometimes prefer it because the parameter spread of past supplies has suited some projects and they were cheaper at the time. Still, I don't see any reason to wait a year for parts when you can buy the same product in stock elsewhere.
Most brands seem to be comparable if not the same on Ft, Cob, Ic and Hfe options for audio drivers in low-medium power applications. However, from earlier datasheets I'd describe On-Semi's product prior to the Fairchild takeover, as a slower device. Perhaps it was intended for general purpose rather than audio use. Currently, I believe we are buying ex-Fairchild product from On-Semi and their datasheet has been revised accordingly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.