Doxa 70 Signature Series Mk2 and BD245C/BD246C

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi! My first post in this excellent forum. A lot of valuable info from many clever enthusiasts and kind people! Thanks! :)

I burned one channel in my Doxa 70 Signature Series Mk2 push-pull amp :mad:, all of the emitter resistors, one PNP, one NPN of 5 in parallel and the PNP of the driver pair and its collector resistor. There is an incorrect schematic circulating on the Internet, which also has been discussed in this forum:

diyAudio

... and also discussed here:

http://www.diyaudio.com/forums/solid-state/142243-doxa-70-clone-2.html#post1803858

I visited the designer a couple of days ago, who happens to live in my town. He told me these transistors went out of production many years ago. He was sorry not being able to help me. He used to order thousand of TI's BD245C/BD246C and sorted them in groups. He's stock went empty several years ago.

I'm aware of other brands and replacements. I'm just worried that the excellent sound in this amp will degrade, replacing the burned transistors. The designer told me that there is no feedback in this amp, making it even more important to match the transistors. Somebody probably will sugest replacing all the transistors in both channels with new replacement. Maybe, but Lauvland (the designer) have no experience with other then the TI transistors. This amp is designed by listening, not by reading transistor datasheets.

That's why i ordered Locki_Z's (http://www.diyaudio.com/forums/vend...rve-tracer-3-0-release.html?highlight=locky_z) curve tracer. My hope is that I eventually get matching curves, by ordering transistors from different sources. Looks like they are hard to find. Looks like most sources are chinese.

The resistors shouldn't cause any problems. This old amp has 5% emitter resistors. The tolerance is not critical, according to Lauvland.

I would really like to get any advice from you! Do you think I will succeed following this route? I really love the sound of this amp, and cannot stand the thought, that I never will get it back.

Best regards
Bjørn
 
Last edited:
--your assumptions are wrong ( type of TRS versus listening impressions )
--you need to have a better understanding of how amplifiers work
--The transistors you are talking about are rubbish
--you are mixing listening impressions with technical analysis ( very wrong )

In reality
Only European designers used/use so slow transistors , that is valid since 1968 , So the perceived result might be nice to your ears but in reality and in real amplifier designing those type of transistors have been abandoned many many years ago ....

Option one
Look in the market for any transistor of today that is close to original specs ,from a trusted manufacturer of your choice ...Local store or ebay is not an option
Buy as many as you can , sample first , match closely , replace all , never mix match old and new parts and you will come up very very close to the original result

Option 2
Get real semis , rugged and fast enough for a decent audio amplifier ,understand what compensation is and does ,alter accordingly ,invest money to be able to evaluate an amplifier NOT the transistors that its made off and expect as a result to have a real amplifier in your hands ...

Kind regards
Sakis
 
Thank you very much for your advices Sakis :)

The reason I'm concerned about using replacements, is this commment:

Do NOT replace the original BD254 and 246 with "better" japan versions!!!
Keep the originals!!
Remember that the constructor of DOXA, Knut Lauvland, have done a lot of testing to find the best possible transistors for the amp. And this does not always meen the ones with the best spec.

In early versions of the DOXA amps, Lauvland used fast and linear transistors like 2sA1227 and 2SC2987. But these were changed to BD245 and 246 by the ear-test!!

http://www.diyaudio.com/forums/parts/54012-doxa-70-se-amplifier-2.html#post847727

... and

Hi, I`m new. I must say a few words. Five transistors are in paralel and one transistor is driver for this five. If you need change only driver transistor you don`t need change and other five. Don`t put any other transistor except BD or TIP. I listen this amp before seven years and I never hear better amp before.

http://www.diyaudio.com/forums/parts/54012-doxa-70-se-amplifier-3.html#post911483

Maybe I should use TIP35C/TIP36C, recommended In the same thread?

Best regards
Bjørn
 
Member
Joined 2010
Paid Member
....Maybe I should use TIP35C/TIP36C...
I believe TIP33/34 are closer, if not equivalent, and both in production by ST and On Semi.
Focus on matching, such as you get with new product supplied in tubes by reputable manufacturers as this will save time and money in learning the hard way - trying to match dissimilar products, batches and losing die sequence in the wafer. 5 devices used in parallel are likely matched only to ensure current sharing, because they will be lightly loaded and running with good linearity for what normally make awful audio transistors.
 
Ahhh ! The Doxa 70. Real wild design , symmetrical "rush cascode" .
https://mrevil.asvachin.eu/amp/topologies/rush/
Actually looked at this one to make another IPS. Notice there is no
compensation whatsoever. The Cob of the (slow) triple EF is the compensation.
Disrupt that with a faster device and there goes stability and SQ.

So the reason for the relatively fast rush and the dog ar$e slow OPS
is to compensate the rush. This is why the original designer stock piled
so many original devices.

I tried running the stock Doxa IPS directly to a modern triple - nope !
Had to first run it through a modern VAS (with compensation) , the
dominant pole had to be set somewhere , and the fast OPS's was too high.

OS
 
Thank you all for much appreciated replies :)

Looks like TIP33C/TIP34C is the way to go. Seem there is consensus that this transistor is nearest match to BD245C/BD246C.

It seems that well known trusted sources have problems delivering the BD's. Looks like the Bournes versions are available to a certain degree. But they are easy to find on places like Shop bd245c online - Buy bd245c for unbeatable low prices on AliExpress.com. You'll find both Bournes and MEV brands. They are cheap, but you never know if they're fake. Still I'm tempted trying some.

I wonder if I should order 20, 40 or even 100 of each kind to get descent matching :confused: Ordering 100 from a source with good reputation easily get more expensive, than buying another second hand amp of the same kind.

Maybe I should order some samples from different sources and check them on the curve tracer, which I hopefully receive in mid July (locky_z's mentioned above).

Best regards
Bjørn
 
Thanks Ian!

I am tempted to try BD samples from China, since they are scarce elsewhere. To me, it looks like the Bournes brand BD's from China, are the same they still offer from good well known sources.

I'm not convinced this will work. They might be fake or far from the original TI spec.

But I'm not convinced that the TIP's will work either. They may also be far from the TI BD spec.

Your link and comment about that look interesting. That might help me. I'll take a closer look at that!

A norwegian member of this forum, kindly offers me 5 pairs of the original TI BD's. The price is steep though. I also think that 5 pairs are to few to get a descent match? One of the drivers (same transistor type) is also burned, so I need a match for PNP/NPN on that one to. I tend to think that I have to replace the transistors in the channel that still works too. Then I need to replace 12 pairs. Don't I need at least twice of that amount or even more, to get a descent match?

Thanks again, Bjørn :)
 
Member
Joined 2010
Paid Member
The problem faced with Chinese copies - or rather substitutes, is as I have tried to show. Despite type marking, the example was a substitute type, not a copy, by Korea Electronics Corp. It has characteristics more like audio transistors such as 2SD1047 and a larger case too. It's really a decent, high gain, high frequency audio type that will almost certainly oscillate in the Doxa circuit. You would need to redesign the amplifier to use such a part.

The quantity needed for matching can't even be guessed, since you don't know what the batches are made up from. They could be sorting rejects or leftover product from factory parts bins, new stock loosely bagged or taken directly from a manufacturers boxing, in sequence. These are very different samples with the chance of a match ranging from very good to very unlikely.

Otherwise, Id stay with plan B and buy a used working amplifier, assuming it had not suffered the same fate in the past. Look after it and don't fit huge or exotic caps or risk shorts, as the design obviously had issues - don't invite trouble.
 
Thanks again Ian :) Very valuable information indeed!

I got new information about the amp (PM), from the guy who knows the amp best here in Norway, apart from Lauvland himself. It turns out that these transistors were not matched in the first place. There is no need, because of the high ohm value (0.5) of the emitter resistors. Buying 5 BD pairs from this guy should be plenty enough.

Then there is the TIP33C/TIP34C alternative, which some of you recommend.

I now feel confident that I will manage to repair the amp and retain the good sound quality :)
 
Thanks again Ian :) Very valuable information indeed!

I got new information about the amp (PM), from the guy who knows the amp best here in Norway, apart from Lauvland himself. It turns out that these transistors were not matched in the first place. There is no need, because of the high ohm value (0.5) of the emitter resistors. Buying 5 BD pairs from this guy should be plenty enough.

Then there is the TIP33C/TIP34C alternative, which some of you recommend.

I now feel confident that I will manage to repair the amp and retain the good sound quality :)

I have to correct myself! I didn't really felt confident what this guy told me about matching, so I called the designer (Knut Øystein Lauvland) himself. Indeed, the resistor values are high and help compensate for the different characteristics among the output transistors. But they were matched to a certain degree by him, though the matching is not critical. As he said, some transistors are far from spec compared to the majority. They were not used.
 
Member
Joined 2010
Paid Member
..... As he said, some transistors are far from spec compared to the majority. They were not used.
This is much more reasonable from a manufacturer's point of view. Nobody should waste their money on products where the value lies in using only a small selection of components.

May I suggest though, that he was using product from the same production batches and that helps a lot for matching anyway. If necessary, you could mix P types from one manufacturer with N types from another but not individual transistors within each set of 5 - they must be as similar as possible before testing and sorting.

You then need to check Hfe for best match (preferably with at least 100mA collector current or more) which can be done quite easily on something like this:
1pc Duoyi DY294 Digital Transistor DC Parameter Tester | eBay
There are very cheap testers for small transistors everywhere but this would be about the cheapest useful instrument that will test real transistors. The toy testers will be useless for power transistors. Note, they get hot with more current and then time becomes a factor, so measure after a timed short period or fix them to a heatsink and attach leads.
 
Thanks again Ian, your help is much appreciated :)

May I suggest though, that he was using product from the same production batches and that helps a lot for matching anyway. If necessary, you could mix P types from one manufacturer with N types from another but not individual transistors within each set of 5 - they must be as similar as possible before testing and sorting.
Yes, that's right! I think he bought 1000 pieces at once.

You then need to check Hfe for best match (preferably with at least 100mA collector current or more) which can be done quite easily on something like this:
I actually got http://www.diyaudio.com/forums/vend...rve-tracer-3-0-release.html?highlight=locky_z at the door today from China. I think it's even better. It's a curve tracer, if you add a PC and software. Tests with current up to 2,5A, if I remember correctly. But your suggestion looks like a cheap and good alternative for hobbyists like me.

There are very cheap testers for small transistors everywhere
Yes, I've ordered one of those cheap component/semiconductor-testers, based on the AVR/AT-Mega-chips. I intend to use that for general component measurements and testing. It's cheap, but not bad. The hardware is simple, because the firmware is complex. A clever design (by a german I think). The chinese are good on taking good ideas and put them into mass production. Got a nice scope from China a couple of days ago, so hopefully I soon have all the proper tools I need for repairing amps.

Note, they get hot with more current and then time becomes a factor, so measure after a timed short period or fix them to a heatsink and attach leads.
Yes, read a lot of warnings about that :) I think that's not a problem with the curve tracer? I've to check that. At the moment, I don't know anything about curve tracers, but I look forward to learn. And it's great to get all the valuable from you and all other clever enthusiasts on this forum :)
 
Last edited:
Member
Joined 2010
Paid Member
The curve tracer is certainly an advanced and complex design but it is only the completed PCB. There is a long way to go and it needs a case, power supply, heatsinks and probably a better off-board connector block before you have a boxed, working instrument - not forgetting to study the operation theory to become familiar with the meaning of results and application. Even so, I'm sure it will be great fun to complete and use. :)

It also tests with quite high current as required to check power transistors fully. At up to 4A DC, it means they will get very hot and certainly need to be attached to a heatsink for curve tracing. It's an amazing device but also requires study, careful technique and good test file record keeping for sensible comparisons.

A professional or very active DIY person would stand to benefit a lot from using a curve tracer but for occasional repairs or builds for the average guy, it will spend most of its life gathering dust. I also have one of Locky-z's modules but it's in the "round tuit" box for another day, as I no longer do amp. repairs ;)
 
Yes, I understand I have to do the homework first, but that's the funny part :)

By the way, I ordered the version ready built, with power, nice alu-box and everything. I admit this unit is not vital for life, but I guess my brother and other family-members also would like to use it some times. My brother has started to set up one of his rooms at home as an electronics lab.

I believe TIP33/34 are closer, if not equivalent, and both in production by ST and On Semi.
Focus on matching, such as you get with new product supplied in tubes by reputable manufacturers as this will save time and money in learning the hard way - trying to match dissimilar products, batches and losing die sequence in the wafer. 5 devices used in parallel are likely matched only to ensure current sharing, because they will be lightly loaded and running with good linearity for what normally make awful audio transistors.
I've searched for defective Doxas and TI BD-transistors here in Norway, but had no luck. It's time to take a closer look at the datasheets for those TIPs. I will probably order a quantity of 100.
 
Looks like the Motorola/On Semiconductor-versions are very close to the BD's. STMicroelectronics might be good, but their datasheet doesn't show any curves. Maybe because they don't want to limit their spec too much. I guess I can't trust these devices? KEC document their version very precisely, but this transistor is far from the BDs and other TIPs.

I found Motorola TIP33C/TIP34C in TO-218 cases at Futurlec. Any experience with them? Will the deliver transistors from the same batch/wafer? My current BDs are TO-218, but I think the TO-247 case will fit as well.
 
Member
Joined 2010
Paid Member
Hi
A wise choice to have the completed curve tracer!

I have dealt with Futurlec and they have a sales office for popular components here in Oz. Based in Thailand, they have a very different sales model to other Asian suppliers. It seems they hold orders over until stocks arrive from their sources so usual delivery time is very :snail:. In my case, a large bag of parts took 3 months (they did warn me when ordering) but all items/quantities arrived as ordered - no fake semis either, as we often fear. TO218 package is obsolete and I have no idea where Futurlec buy and in what packing but they are obviously not authorized stockists if they offer obsolete parts. If you do buy in original packing, you need only buy 1 tube etc. There's no point (perhaps even a loss of sound quality) in trying to micro-match when it's never been required.

My point in showing that obsolete KEC datasheet was that you need to know what the manufacturer's attitude to type markings is. In traditional markets, industry regulations once ensured the type was established by the OEM specifications. 2nd source manufacturers supply similar product with their own branding after it becomes obsolete but you seldom see a full set of performance graphs from them. Many Chinese supplier datasheets I see, are simply copied over from the OEM datasheet and circulated among several manufacturers, the PDFs having each manufacturer's details changed. Testing a few parameters soon sorts out who's kidding.

ST Micro, who supply a lot of excellent second source product, don't give full details either. I think very few second sources do, so I wouldn't be too concerned, since the reputation of the Manufacturer is probably the most important quality. That's why counterfeit product is such a threat, because it damages reputations.

ST also made BD245/6 at an earlier time, concurrent with the TIP series, so it would be naive to think ST would set about making different copies of what were intended to be copies in the first place. The problem will be to find enough of the PNP type TIP34 which may be obsolete everywhere now too - don't delay searching. As I said earlier, the match is required for current sharing among the 5 parallel types but not necessarily between both PNP and NPN types. They are in series where matching cannot be critical, because BD PNP/NPN types are not all that complementary anyway. Yes, disappointing to learn that complementary parts aren't really complementary - just the leading ratings may be. :eek:
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.