As I recall, in the 1970ies the BDY56 power transistor was listed by members of the Philips group, such as Valvo. The label in discussion surely doesn't mean Valvo, but maybe another family member instead.
Sorry for being somewhat offtopic.
If that is of anyone's interest, a friend of mine has got a bagful of these BDY56's - NOS, of course!
Best regards!
Last edited:
I haven't looked at doing a Bode plot. I am working on other stuff at present - I think Naim claimed their amplifiers could safely drive loads to angles of +/-90 degrees.
Laying the cursor over .inc statements and right clicking on abbreviated link should open the .op command dialogue box. The "Open" option there should take you to the relevant .model file at On.Semi which can be copied and pasted into the sim as a comment.
Last edited:
My stock of NOS BDY56's from 1978 also shows this logo. I believe it to be a long since defunct Clevite brand.
Brian.
Member
Joined 2009
Paid Member
I'm away from home for a couple of days so no access to run Spice. But there are four aspects of the amplifiers stability that I believe need proper attention:
a) thermal stability - relying on a closed box and non-contact Vbe device is poor in my view and I've addressed this one.
b) loop stability - not yet satisfied about this until I can get a decent bode plot. I assume any issues can be slugged with Cdom but I believe the sound I'm searching for comes when Cdom and the phase lead compensation are both at the right place. Sticking with the original zetex VAS device suggests the 39pF Cdom is a 'safe bet' but the phase lead across the feedback resistor works more closely in concert with the choice of output devices I would suspect
c) cross-conduction / shoot-through, something the quasi is known for where lingering minority charge carriers in the base of the output devices prevents them from turning off when desired and at high frequencies you get some 'Class A action'. I believe this issue is why NAIM wanted people to use their pre-amps, so that ultrasonic frequencies were excluded and over-heating of the output devices due to shoot-through is minimized.
d) load instability - lack of an output inductor requires NAIM recommended speaker cables. I don't like this idea, I've included the provision for an output inductor. I'm not yet convinced this is detrimental the sound in any way. Opinions ?
The question of the phase networks on the driver base-stoppers is still open. I'll see if I can dive into it with Spice.
The pcb vendor says my Gerber files are fine so once Chinese New Year is over they'll process them.
a) thermal stability - relying on a closed box and non-contact Vbe device is poor in my view and I've addressed this one.
b) loop stability - not yet satisfied about this until I can get a decent bode plot. I assume any issues can be slugged with Cdom but I believe the sound I'm searching for comes when Cdom and the phase lead compensation are both at the right place. Sticking with the original zetex VAS device suggests the 39pF Cdom is a 'safe bet' but the phase lead across the feedback resistor works more closely in concert with the choice of output devices I would suspect
c) cross-conduction / shoot-through, something the quasi is known for where lingering minority charge carriers in the base of the output devices prevents them from turning off when desired and at high frequencies you get some 'Class A action'. I believe this issue is why NAIM wanted people to use their pre-amps, so that ultrasonic frequencies were excluded and over-heating of the output devices due to shoot-through is minimized.
d) load instability - lack of an output inductor requires NAIM recommended speaker cables. I don't like this idea, I've included the provision for an output inductor. I'm not yet convinced this is detrimental the sound in any way. Opinions ?
The question of the phase networks on the driver base-stoppers is still open. I'll see if I can dive into it with Spice.
The pcb vendor says my Gerber files are fine so once Chinese New Year is over they'll process them.
Last edited:
The SPICE model for MPSA56 doesn't thrill me on VAF much although the BF is high - there are other Motorola transistors I can look up
The Fairchild version of this is slightly different and has a better VAF. Looking back through an old Farnell catalogue I see that Zetex manufactured these in their E-line package - I could not track down their SPICE model as production of this has ceased.
Off to a good start.
Some more suggestions for your PCB design:
- You may want flexibility to swap parts in and out easily when investigating what affects the sound quality and why. Consider transistor sockets. Eg: the LTP is one area you may want to try parts until you get minimal output offset dc.
- When I put the star ground on my PCBs I allow for an M4 bolt hole so I can bolt on cables from speaker gnd, psu cap gnds, chassis earth, etc. These wires can stack with eyelets and be bolted down using a shake-proof washer.
- I am suspicious of the linearity of those output FETs. I do not know, tho! But in the face of ignorance I would choose a high quality relay.
- If you aren't using an output over-current protection circuit, consider an output dc voltage detector. When power transistors fail they either go open circuit or close circuit. The latter will destroy your speaker. I know from bitter experience! The detector can latch your relay or FETs. Some circuits use a triac to short the output but this is a bit crude.
- Using bypassable safety resistors in the supplies is a good idea while you are getting it working. I usually use 1-ohm 5W or thereabouts. Buy a few in case you melt one.
- Consider adding probe points. Could just be a hole to stick a short wire post in. Sometimes you need to measure something and, for instance, clipping a probe to a transistor lead can be tricky and scary when the thing is powered up.
- Don't forget to put your feedback dc blocking cap the correct way around.
Some more suggestions for your PCB design:
- You may want flexibility to swap parts in and out easily when investigating what affects the sound quality and why. Consider transistor sockets. Eg: the LTP is one area you may want to try parts until you get minimal output offset dc.
- When I put the star ground on my PCBs I allow for an M4 bolt hole so I can bolt on cables from speaker gnd, psu cap gnds, chassis earth, etc. These wires can stack with eyelets and be bolted down using a shake-proof washer.
- I am suspicious of the linearity of those output FETs. I do not know, tho! But in the face of ignorance I would choose a high quality relay.
- If you aren't using an output over-current protection circuit, consider an output dc voltage detector. When power transistors fail they either go open circuit or close circuit. The latter will destroy your speaker. I know from bitter experience! The detector can latch your relay or FETs. Some circuits use a triac to short the output but this is a bit crude.
- Using bypassable safety resistors in the supplies is a good idea while you are getting it working. I usually use 1-ohm 5W or thereabouts. Buy a few in case you melt one.
- Consider adding probe points. Could just be a hole to stick a short wire post in. Sometimes you need to measure something and, for instance, clipping a probe to a transistor lead can be tricky and scary when the thing is powered up.
- Don't forget to put your feedback dc blocking cap the correct way around.
Member
Joined 2009
Paid Member
Generally, the saturation resistance of FETs is very linear, depending for the most part on a stretch of intrinsic silicon. But audiophile ears can be very discerning so I'll take the recommendations of earlier posts to choose a FET with very low resistance thus minimizing the signal voltage developed across it. One good choice might be FDP083N15A. It is more than an order of magnitude lower in resistance than the device I showed on my original schematic.
I have read many people believe that mechanical relays have the potential to be even more damaging to the sound. I'm not sure if this is true but I am convinced that they can easily fail to break a high current dc signal due to arcing between the contacts. This risk can be reduced by using the catch diodes on the output and switching the speaker to ground with a SPDT relay but I have opted for the SS relay as being a safer bet as well as being more compact.
If you want to see how many hours and engineering efforts went into the analysis and development of SS Relays here, take a look at Bonsai's long thread that began with relay failure: http://www.diyaudio.com/forums/solid-state/191449-output-relays.html
and here: Solid State Relay with PCB Layout
Thanks for the links. So far I have read about 13 pages of the first one and glanced at the second.
Clearly a SSR is belt and braces. The question of distortion was not objectively addressed which is crucial. There are ways to mitigate relay arcing using caps and snubbers. However, the case of a temporary dead short on the output is problematical in the absence of some current limiting mechanism.
Clearly a SSR is belt and braces. The question of distortion was not objectively addressed which is crucial. There are ways to mitigate relay arcing using caps and snubbers. However, the case of a temporary dead short on the output is problematical in the absence of some current limiting mechanism.
Last edited:
Member
Joined 2009
Paid Member
To get the optimum inductance in a set of Naim leads one needs two five metre lengths of their special cable inconvenient in terms of cost and stiffness. An output inductor can save one some trouble in this regard.
Cross conduction can happen to any amplifier that has marginal stability. An h.f. incipient signal at the inverting input can increase the gain at h.f. despite an audio signal stimulus at the non-inverting input requiring a reduction in output.
There is a passive filter at the input of Naim power amplifiers - the bandwidth limiting is handled at this point. I think they put the filters for CD in their players these days but there was a time when these were in their pre-amplifiers.
I looked at the phase correction networks in series with the drivers. The plots for the complementary feedback pair and combined output of both halves are almost identical.
The Darlington upper pair are almost the same except these have a higher voltage ceiling at the top. It is possible to reduce conduction in transistors by putting a voltage on the emitter that causes the base voltage to reduce. This is what results at the output summing point - not too much different in principle from what happens in an LTP stage.
Because the output halves sum in phase and the transfer is clean the amplifier can be run close to Class B - not unlike the Quad 303.
In the simulation I posted, the standing current in the output stage is 3-4 m.a. the bulk of which is supplied by the driver transistors. These are mounted on decent heat -sinks.
If the driver transistors can run cool as the bias transistor then this need not be mounted elsewhere to sense heat.
If the standing current in the output stage is to be increased then this thinking will need to be reviewed.
To that end a good many designers arrange the complement of output and driver transistors and the Vbe heat sense one to be on a common heat-sink.
In an amplifier of my own design I did not fit the driver transistors to the main heat-sink and had trouble controlling the current.
I had glued the TO-92 Vbe transistor on the metal case of one of the outputs as tried by Self. The outcome was that the glue melted.
The NPN driver is a BD139 and so I replaced the TO-92 Vbe with another BD139 and bolted it piggy-back style onto the NPN driver. That fixed it.
It is more than 20 years since I built this -the standing current is modest but still quite a lot more than the Naim simulation.
As the NPN pair is more affected by heat in Quasi comp - making the NPN driver the Vbe sensing point is an option you might consider.
If the aim is to make an authentic Naim then using an output resistor rather than an inductor would seem appropriate. If you intend to use high C speaker cable or speakers then you could add a series inductor (so you are faking the cable rather than changing the amp). And put it outside the case.
There is a reason JV did not use an output inductor. There is always a reason.
There is a reason JV did not use an output inductor. There is always a reason.
How many DIYers have taken the simple step of adding a resistor to their speaker cable to see what difference it makes?
Member
Joined 2009
Paid Member
I won't have any special NAIM cables so I have put space on the pcb for the inductor and I do feel safer using one but it will be relatively easy to try without. I don't have electrostatic speakers and I don't believe my cables are especially capacitive - at least not the cables I plan to use.
There are a few things that JV did with this amplifier that were constrained by his choice of speaker. He was developing / using electrostatic speakers. He had trouble with and ongoing risk with these speakers producing 'shorts' for the amplifier which is why the protection circuit was so important back then. Few people will want to use these single slope SOAR protection circuits on a modern clone.
He also had relatively 'slow' output devices and would have been more worried about cross-conduction and the need to restrict the upstream system bandwidth.
It's not clear to me what needs to be included in my clone in terms of 'sound' vs. what was needed for the 1970's - if you have further guidance on part choices etc. that would be good. The amp will be used with 'modern' output devices though, 'modern' cables and 'regular' speakers.
Last edited:
I think, with the BDY56's, as pictured in a former posting within this thread, he took the best he could get. A transition frequency of 10 MHz wasn't too bad in those days. For what I have the oversight, these BDY56's were the fastest devices of that power then, faster by far than most of their contemporaries.
Best regards!
There's also the need to protect against a customer shorting the speaker cables while connecting them.
What semiconductors do you have in your top drawer already? And how many of each?
Member
Joined 2009
Paid Member
For this project I was planning to use Fairchild FJL5215/4315 which are close equivalents to the Toshiba 2SC5200/2SA1493. However, on checking, I have only 3 complementary pairs and I'll need 4 of the n-type. I have more than a dozen complementary pairs of the Toshiba parts but all in TO-220. It's no big deal to order some parts from Digi-key; they're $3-64 each. They also stock Sanken 2SC3263 for $5-71.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.