"Quite possibly" - are you making a hint because you have seen this part or is this a wild guess ?
I would not have looked at this had you not pointed to the change in the Vas transistor.
I could not identify the part from first hand knowledge but sometimes one can glean some more information by closer studies and so I looked at the images for the NAP 110 and NAP attached to posts 3 and 19 to see what else there was there.
See if you can see what I have seen, and why I reached the conclusion I did.
Member
Joined 2009
Paid Member
I can only say that it appears NAIM made the change when the increased the power rating of the amplifier. This means higher rail voltage and higher power dissipation in the VAS. At lower voltages you also want lower Vce(sat) - related to base width modulation. But even if this is the case, I could not have surmized this part number ?
On a close look, I think we'll find that the small signal transistors are sitting on the round plastic ferrules that were popular for high quality builds back in 1970s. That may be confusing anyone trying to identify the transistors by their outline.
The point being that those are E-line transistors fitted as VAS transistors, just like other models. This a critical parts issue, because of the unusually high capacitance of the types used not only by Naim but on serious clones where the penny has dropped as to where the secret sauce begins. I haven't seen all models and versions by any means but ever since my first service work back in the 1980s on a NAP120, I've only a few times seen anything different in the semis lineup than Motorola small signal types, MJE243/253 drivers plus Ferranti/Zetex E-line VAS transistors.
NAP 90, the precursor to the smaller NAIT models, is an exception at the output stage, due to much lower power requirements.
You might also notice the stylized '70s PCB copper pattern - hand drawn by a commercial artist, I'd say. It didn't make any difference though, when it changed on later models to straight runs and elbows of fixed track width, as you got using adhesive tape, hand applied artwork or perhaps a crude software program of the day, though I don't know of such in audio electronics before the IBM PC arrived. The copper pattern seen on the excellent NAP200 and clones though, is in the modern league with all the bells and whistles of advanced CAD layout.
IMHO, if there's any real secret in the PCB layout, it's in the late models and revisions. The early model artwork is too inconsistent to call a special design element and if the reasons given held true at audio frequencies, it would have been time to quit electronics and take up gardening.
Member
Joined 2009
Paid Member
No, I really don't think so. Look again a the NAP 110 VAS device. It's a TO-92 alright.
Attachments
The image attached to post 3 is an early version of the NAP for sure - apart from the plastic ferrule transistor mounts the low wattage resistors are carbon film. The Vas transistor in this image has a different lead configuration of ecb which marks it out from other BC or 2N types. If the usual brand for transistors was Motorola it seems entirely possible that so was the Vas transistor and Motorola were making BC640 as part of their component range.
I managed to find another image depicting a later version of the NAP110 naim power amplifier nap 110 2 naim - Hifi audio In this the ZTX transistor has replaced the ebc lead out
TO-92 type, the output transistors have been given Naim part mumbers and the low wattage resistors are film.
Some research shows Julian Vereker designed amplifiers up to 1989 and Roy George took over this role. There are those who think it matters who the designer was because of subtle differences in sound-somewhere some other time I read it was acknowledged this could happen between production runs of the same model.
Half the fun in DIY is trying things out to determine a preference. There was some discussion on a Naim forum where there was a preference for an NAP110 over a NAP140.
I have posted a copy of the SPICE file for BC640 here is the ZTX753
*ZETEX ZTX753 Spice Model v1.0 Last Revised 24/5/93
*
.MODEL ZTX753 PNP IS =3.2007E-13 NF =1.0041 BF =200 IKF=1.6 VAF=76
+ ISE=8E-14 NE =1.57 NR =1.0008 BR =33 IKR=0.45 VAR=51
+ ISC=6E-14 NC =1.079 RB =0.087 RE =0.08 RC =0.07
+ CJC=80E-12 MJC=0.4896 VJC=0.7676 CJE=350E-12
+ TF =0.86E-9 TR =24E-9
*
*$
*
It does a little better than the BC640 in current gain (BF=200) but less well in Early Voltage (VAF=76).
A point to consider here is if the early voltage is higher with BC640 what advantage is there in having a greater collector to base capacitance of the ZTX753.
Hmmm. Maybe I should make the time to do a simulation.
I can't argue with the TO92 VAS transistor package. Let's find a TO92 type from the early 1970s that would be rated high enough (80V) and able to dissipate the best part of 1W to handle the duty comfortably. BC640 might seem feasible but no, I don't think they were around then and I doubt Mullard/Philips would have been approached to supply.
What's more, just try substituting them. Like BD140, it makes the amplifier sound quite different and essentially wrong compared to ZTX753 and of course, even lower Cob types like 2N5550/5400. Yes, tried them several times on different models with consistently poorer sound quality in my estimation, at least.
Sure, there was then a wider range of BJT parts but one can also re-establish the total Miller capacitance required for stability by increasing Cdom. SQ still suffers though so I still say that the VAS transistor's Cob value would or should have been much the same as the type normally specified for other NAP designs. I believe that high Cob VAS transistors were found desirable because the transistor then acts with more of the the character of a varicap when signal is applied and it produces a small sound effect as a specific type of distortion, rather than having any accepted technical advantage. Lose that sound and you lose much of the the Naim sound character
Looking again as you suggested, I see the VAS transistor is not balanced by another TO92 transistor as its CS. That is of little consequence but it seems odd to me - perhaps out of character for the designer or there's no complement for a PNP type? If you look at the beefed-up 110 version, NAP140, you can see that though the supplies and circuit are the same, the usual E-line VAS type is fitted and there are no pretty touches by a restorer -just basic hand assembly.
There have been some changes on the 110 boards though:
I don't recall stacked film caps or so many tiny polystyrene caps used back in the 1970s. There has been a problem with one WW resistor overheating whilst others have just been been replaced, not that it matters much. The output transistors are of eastern European (Hungary?) NOS origin and from more recent times, I think.
I admit to not looking closely enough to check the circuit under the parts because I tend to make even more errors with that but I have worked on a couple of NAP110s in the past. I will cast nasturtiums on the authenticity of components and extra touches you see in that pic, though.
What's more, just try substituting them. Like BD140, it makes the amplifier sound quite different and essentially wrong compared to ZTX753 and of course, even lower Cob types like 2N5550/5400. Yes, tried them several times on different models with consistently poorer sound quality in my estimation, at least.
Sure, there was then a wider range of BJT parts but one can also re-establish the total Miller capacitance required for stability by increasing Cdom. SQ still suffers though so I still say that the VAS transistor's Cob value would or should have been much the same as the type normally specified for other NAP designs. I believe that high Cob VAS transistors were found desirable because the transistor then acts with more of the the character of a varicap when signal is applied and it produces a small sound effect as a specific type of distortion, rather than having any accepted technical advantage. Lose that sound and you lose much of the the Naim sound character
Looking again as you suggested, I see the VAS transistor is not balanced by another TO92 transistor as its CS. That is of little consequence but it seems odd to me - perhaps out of character for the designer or there's no complement for a PNP type? If you look at the beefed-up 110 version, NAP140, you can see that though the supplies and circuit are the same, the usual E-line VAS type is fitted and there are no pretty touches by a restorer -just basic hand assembly.
There have been some changes on the 110 boards though:
I don't recall stacked film caps or so many tiny polystyrene caps used back in the 1970s. There has been a problem with one WW resistor overheating whilst others have just been been replaced, not that it matters much. The output transistors are of eastern European (Hungary?) NOS origin and from more recent times, I think.
I admit to not looking closely enough to check the circuit under the parts because I tend to make even more errors with that but I have worked on a couple of NAP110s in the past. I will cast nasturtiums on the authenticity of components and extra touches you see in that pic, though.
Attachments
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
I have managed to expand the image a bit better on a Tablet and the lead profile of the transistor fitted appears to be ebc (front face view) L14 or L22 which would also suit a ZTX753 - why not use that.
If correct then this excludes BC640 and the BFR's and opens the possibility of a device such as MPSA56 - this is rated 625mW there is a little more pcb copper attached to the collector (heat dissipation).
An extra tad of capacitance in the traces might make it a usable replacement for the ZTX. I see using the pcb as part of the circuit as a form of copyright protection.
Anyway the gerbers are off to be made up so it is a case of waiting to see what happens next.
Member
Joined 2009
Paid Member
I'd like to use the ZTX for the VAS and I'm equally happy choosing it's complement for the CCS device. I put as much copper around the collector pins of both devices as I could within the very limited space I have on this compact board. But I anticipate gluing a small piece of metal to the transistor package on both devices to extend the heat radiation areas, i.e. provide a kind of 'cooling fin'.
Almost certainly. Nice detective work and interesting comments on TR4 identity lohk and mjona. Never a dull moment here
Member
Joined 2009
Paid Member
Almost certainly. Nice detective work and interesting comments on TR4 identity lohk and mjona. Never a dull moment here
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 - we are having the stormiest and wettest summer here in a very long time. The grass is getting longer here and it will be the same at our holiday house up the coast where the lawn normally goes brown in summer.
Last edited:
Member
Joined 2009
Paid Member
You mentioned a few strange things at high frequency in your simulation.
Rise and fall times of a 10kHz square wave are representative of higher frequencies and I have strung a load of 8.2ohms with 2uF in parallel to see how good the settling time is and if there are any ripples or overshoots - there are none - but I have not built anything based on simulations yet - sometimes small amounts of capacitance can be more troublesome than a larger amount. I will leave that for others curious enough to investigate.
I have just done a basic version of the circuit - I found a problem arose with MJE253 in the circuit so I chose BD140 and BD139 for the driver role - These were in the model I adapted for an amplifier of my own design with complementary symmetry.
The Naim simulation works just as well with Vas as 2N5401 which is available in the LT SPICE library - so that can be used if there is a problem with the ZTX753 file. The diodes are all taken from the SPICE library - the output spike protection ones are overkill but fast recovery ones are better than standard.
I have set the standing current fairly low - R20 can be changed to suit. Similarly R11 can be changed to set the output dc voltage.
I have not taken the time to sort the parts designations out - if other people find a use for this they may want to add other features as you have done.
After days of lashings of rain and wind the sun has come out - fingers crossed that it lasts.
Member
Joined 2009
Paid Member
Thanks mjona - I'll download and check this out when I get back from the office later today.
I also did a Spice simulation (to work out a fix for the temperature compensation and to explore the PSRR performance). The issue was when I tried to implement a bode plot of the feedback loop by inserting a probe into the feedback loop. I've used this approach before with good results. This time it didn't produce what I was expecting. So as an alternative I also ran a square wave and it looked great but I prefer to see a full bode plot that makes sense. Most likely I wasn't doing something right.
I also did a Spice simulation (to work out a fix for the temperature compensation and to explore the PSRR performance). The issue was when I tried to implement a bode plot of the feedback loop by inserting a probe into the feedback loop. I've used this approach before with good results. This time it didn't produce what I was expecting. So as an alternative I also ran a square wave and it looked great but I prefer to see a full bode plot that makes sense. Most likely I wasn't doing something right.
Member
Joined 2009
Paid Member
Kay - sorry, I'm not familiar enough with the manufacturers labels.
Mjona - I still need to download the device model from on-semi you used in order to make your simulation work. Unfortunately I have some day-job priorities to deal with so I'll have to come back to this in a couple of days. Have you tried making a bode plot for the feedback phase margin ?
Mjona - I still need to download the device model from on-semi you used in order to make your simulation work. Unfortunately I have some day-job priorities to deal with so I'll have to come back to this in a couple of days. Have you tried making a bode plot for the feedback phase margin ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.