NAD 7020i sudden death

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

PRR

Member
Joined 2003
Paid Member

Attachments

  • 2N3055-1966-42.gif
    2N3055-1966-42.gif
    59.3 KB · Views: 219
  • 2N3773-1966-42.gif
    2N3773-1966-42.gif
    61.6 KB · Views: 215
> The RCA hometaxial was rated at 150W

RCA 1966 (hometaxial) 2N3055 was rated 115W.

This part shows ~~1.3Vbe at 5 Amps Ic, so "acts like" 0.14 ohms internal emitter resistor due to significant parasitics.

2N3771/3 was rated 150W.

You are right, that figure was quoted in another source. I should have checked this in the RCA Handbook. There is no MJ2955 hometaxial complementary device in this Handbook.
 
In post 17 I suggest .47 ohm emitter resistors. I used .51 in my AX6. There are successful epitaxial transistor amps with .33 ohm, .22 ohm, and possibly even .1 ohm emitter resistors.
]
Keeping the OT wires 1" away from each other, and keeping the temperature sense wires 2" away from output transistor wires, are key IMHO avoiding oscillation. You'll see my base & emitter wires were 4" long so I could flip the driver board over for work on each side. The high current stuff is IMHO, not sensitive to position or length inside the RF free case.

Note nigel7557 or whatever has had trouble with oscillation in Maples (?) 1970's amps that came with homotaxial transistors. His solution was to put 10 ohm 1 watt resistors in the base line to the output transistors. I did that as a damage limitation during meltdown practice, when I replaced the 1970 RCA transistors (burned ) with NTE60's (MJ15003 white box?) epitaxials in my dynakit ST120. I replaced the wires from the driver board to the bases of the transistors over on the heat sink, with 10 ohm 1 watt resistors. Flying right out in the air. If the OP gets oscillation, I suggest that is the next step. As I suggested in post 14.

Replacing the RCA devices with MJ15003 was the key move which increased the stability margin in your amplifier.

The MJs have higher fT than the RCAs which means that decline in gain and the point where phase shift starts is moved to a higher frequency.

Your amplifier is compensated for the lower frequency point where phase shift starts with RCAs.

I would not rule out aging of components for their burn out.

On radiated emissions consider the primary of an ignition coil where the current is turned on and off. You have the same effect with wire wound resistors and variations in current passing through these.

You can check out from other sources what I said about running collector and emitter leads closely to cancel these effects. If you space these out there is a wider radiation field inside your chassis.
 
Replacing the RCA devices with MJ15003 was the key move which increased the stability margin in your amplifier.
The MJs have higher fT than the RCAs which means that decline in gain and the point where phase shift starts is moved to a higher frequency.
Stability of the ST120 was not a problem. Use of just enough aluminum heat sink to pass the FTC 1 hours 60w/ch test was. Lots of comments about "the worst amplifier ever sold" out there. Also lack of output transistor idle bias current when the music was soft caused crossover distortion. The HD tests are at full power, so this didn't show up in the advertising. Just in the terrible magazine reviews. 200 khz Ft output transistors on 1966 build didn't help high frequency IM distortion. My use of MJ15003 solved that problem. Dynaco went to TIP3055 or 2n3773 in 1970.
I was lucky that the MJ15003 worked with no emitter resistor on the lower transistor, although the idle bias current was set by measuring the current through the .51 ohm resistor between upper OT emitter & lower collector. When the PC15 driver became unrepairable due to patches caused by unavailability of TO5 drivers, an AX6 board (point to point) took over the job with emitter resistors on both upper & lower output transistors. Both PC15 side with djoffe closed loop idle bias control modification (7 transistors), and AX6 side, sound great 1/8 watt or 72. Two PCAT fans solve the heat problem.
Spacing the OT b & e leads 1" apart works great. Radiated emissions are more a problem from 1000 w CB radios driving by emitting barking dogs performing Dixie or an AM radio station. Solved by changing input cap on AX6 from 47 pf to 82. Unit already had a output coil, wound around the output capacitors. PC15 side had 150 pf input cap.
 
Last edited:
Facing some Facts

You say the output transistors used in the Dynaco ST120 output did not help IMD - officially quoted as less than 0.5% the same as for THD.

The 1978 RCA Power Devices handbook gives fT as 800kHz rather than the 200kHz you quoted. The latter figure would align with the RCA 2N3773.

If we accept 800kHz for the Dynaco and ditto for the NAD 3020 "speculative" hometaxial output transistors, the constraints on the gain bandwidth level and distortion results would be similarly affected.

That is not the case when one compares the specifications attached for the NAD 7020 which speak for themselves.

I have searched for detail in a NAD 3020 youtube re the maker and date codes of the output transistor pairs - these RCA devices described earlier as hometaxial.

The giveaway to this theory is the devices have 1985 date codes. If anyone wants to check that see Vintage NAD Amp teardown - YouTube

There is a hometaxial 2N3055 in the 1978 RCA Power Devices Handbook which recommended for Quasi Complementary Output stages.

That there is no hometaxial PNP of any description to match this in the handbook presents as an inconvenient fact which people choose to overlook.
 

Attachments

  • NAD7020spec.PNG
    NAD7020spec.PNG
    109.8 KB · Views: 193
  • NAD3020.jpg
    NAD3020.jpg
    110.8 KB · Views: 196
Last edited:
The NAD might have been built in 1978 or later.
If we accept 800kHz for the Dynaco and ditto for the NAD 3020 "speculative" hometaxial output transistors, the constraints on the gain bandwidth level and distortion results would be similarly affected.
My ST120 was built in 1970, going by date codes. The original units in magazine reviews were sold in 1966. A lot of progress was made in power transistors 1966-1978.
In RCA 1972 databook for power transistors there are no TO3 pnp transistors at all. All the homotaxials are npn.
 
Last edited:
I noted your comment about the lack of an emitter resistor in the lower NPN output transistor.

It is more usual in conventional designs to leave that out and put a suitable value in series with the collector where the vbe will not be reduced by local feedback.
Dynaco have not followed that approach either.

The upper NPN has an emitter resistor connecting to multiple points where local feedback signals will arise and putting aside the usual thinking of current flowing from positive to earth I looked look at paths for the charges which flow from in the opposite direction. These are capable of changing the base emitter voltage of the upper NPN transistor and act as controls.

One of these paths is the 7 Watt 300 Ohm resistor between the upper NPN transistor collector and earth which is a clean source of electrons and the voltage drop across the resistor could be regarded as linear.

To me this kind of use of local feedback is an interesting approach.
 
Last edited:
Dynaco have not followed that approach either.
The upper NPN has an emitter resistor connecting to multiple points where local feedback signals will arise and putting aside the usual thinking of current flowing from positive to earth I looked look at paths for the charges which flow from in the opposite direction. These are capable of changing the base emitter voltage of the upper NPN transistor and act as controls.One of these paths is the 7 Watt 300 Ohm resistor between the upper NPN transistor collector and earth which is a clean source of electrons and the voltage drop across the resistor could be regarded as linear.
That dingbat $1 300 ohm 7 watt resistor and the $.75 non-polar cap it feeds are the problem on dynaco ST120. There is minimal idle bias current on output transistors at low wattage until louder output charges up the capacitor. The djoffe mod uses 7 transistors to measure the idle current across the .51 ohm emitter and force 20 ma. Dynaco Stereo 120...can be beautiful - diyAudio
I built two and the mod forced the dynaco design to sound great at all wattage. But the add on has a tendency to blow the sense transistor emitter junction & go 200 ma idle current. Fortunately my fans blew off the excess heat without output transistor damage. The AX6 dispenses with all that and sounds good with only 6 transistors 18 resistors and 6 e-caps. Retro Amp 50W Single Supply - Page 22 - diyAudio
Dynaco should have cut up more Armstrong 621 or Leak delta 70. Which came first? ST120 was 1966.
One dynaco board lost too many lands due to damage from replacing TO5 drivers with TO220 drivers. So I built a point to point AX6 and replaced it. The AX6 pcb artwork won't fit in the 3.5" tall ST120 case. Other than a slight gain difference covered by adjusting the balance pot on the preamp, the two channels sound the same - as good as my CS800s at 1/8-70 watts.
To your point about no homotaxial pnp, you're welcome to buy an NAD7020 and repair it without emitter resistors. As many times as necessary. Mr. Finch and others advocate that epitaxial output transistors last longer with emitter resistors. Also they say the idle bias current is easier to set right with emitter resistors. I'm not buying a NAD, ebay NAD's here cost 5 to 310 times what Peavey products cost with big honking .5 ohm emitter resistors OEM. Listening to a MMA-875t now, cost $35, monaural with speaker safe transformer output. An actual stereo M-2600 came in from ebay last week, $105, also works and sounds as good as these **** $3 speakers I'm having to make do with can show.
 
Last edited:
wow... it seems I stirred quite a debate :)
(too bad I can't make much of it - yet, at least)

@zenzaman

If you still have the defunct power transistors could you post an image or otherwise tell us what brand these are and the full part numbers.

yea, I added a picture of them.

as for emitter resistors I found myself in a bit of a 'real estate shortage' on the board so found a good workaround (as the main heatsink already had M4 and M3 holes drilled in it). I etched a small PCB, soldered the resistors to it, bolted the PCB to the heatsink and then routed wires to the backside of the main PCB, cut the traces and soldered the wires. I didn't play any music yet (it's 2AM here), nor measured DC offset or idle current again, but I did plug it in and the DBT showed no misbehaving.
I only had .33 3W and .56 5W resistors and went for the .56 but I can always swap them for a different value if needed (and recommended)
 

Attachments

  • IMG_20210122_234024.jpg
    IMG_20210122_234024.jpg
    361.4 KB · Views: 117
  • IMG_20210122_223215.jpg
    IMG_20210122_223215.jpg
    366.3 KB · Views: 87
  • IMG_20210122_225153.jpg
    IMG_20210122_225153.jpg
    255.5 KB · Views: 85
  • IMG_20210122_232625.jpg
    IMG_20210122_232625.jpg
    341.5 KB · Views: 81
  • IMG_20210123_011604.jpg
    IMG_20210123_011604.jpg
    628.2 KB · Views: 89
The debate is an extension of much the same as a thread last year only with fire breathing in my direction.

Anyway that caused me to do more searching to find this that thread - see Hometaxial 2n3055/mj2955 and read post 2 and the work background of the member responsible.

I have looked at a variety of NAD amplifier models which omitted the emitter resistors and these were produced over two decades and selling on the second hand market.

I have a contact who runs a shop selling used Hi-Fi equipment I will ask him to keep a lookout for me on any model coming into his shop. These are not selling for silly money in this country - maybe owners are mostly hanging on to them.
 
I had amicable contact with John Ellis two years ago. John had started a thread for a new amplifier circuit he had built. He was using other than LTSpice software for simulations. I was able to do those in LTSpice for him.

Anyway there were two points in post #5.

You have skipped John's first one which supported my contention that the 7020 output transistors are epitaxial devices. In a way I have already acknowledge his expertise on that score.

The outputs that came out of zenzaman's amplifier are clearly Motorola 2N3055 and MJ2955 which are epitaxial according to datasheets and the Motorola Selection Guide

The year code in the images give the year code as 1990 so they were not old stock at the time of manufacture and the NAD 7020 company part numbers end with M.

What have you got to say about this.
 
The higher the emitter resistor value the higher the local negative feedback signal will be and THD will also be affected because this will cause the output and driver in each output have to have reductions in base emitter voltages.

You will then have to re-bias the output stage. The ideal set up voltage drop across an emitter resistor is 26mV at room temperature which is 25 degrees C.

A scientist named Barney Oliver who worked for Hewlett Packard wrote a paper explaining what is involved. You should be able to track it down by doing a search on the web.

I have never built a Class AB amplifier without emitter resistors. The trouble with these that silicon becomes more conductive with temperature so it is a job to keep the bias tracking as it should.

On Semi invented Thermatrak transistors with sense diodes on inside the die and there is an Application Note showing the performance of these that you can download but even these are not perfect.
 
also, are the 0.56ohm resistors ok in value or too much? (I didn't go for the 0.33 3W I had because I have no idea what make they are. If a smaller value is needed I might be able to find a couple from the same local seller that had these .56 ones)
You'll likely have to adjust the bias pot to get 20 ma idle bias current. 20 ma across .56 ohms is 11 millivolts. Looks like pot R444 may do the job but possibly a pot parallel to R442 may be necessary. Possibly a higher value of R442 may be necessary. Put two clip leads on one of the emitter resistors and measure this voltage cold and hot. You can speed up hot with a hair dryer.
Higher emitter resistor values cut peak watts slightly. Since the ear responds logrithmically, this is practically inaudible at the top of the volume range. My ST120 pries 72 w/ch out of a single pair of output transistors with .51 ohm emitter resistors, but that is with 70 v rail and a 437 VA transformer. Your 7020 appears to have +-30 v rails which adds up to 60 v. Bargain products may have transformers that sag in voltage more near the maximum wattage.
10 ohm base stopper resistors on the output transistors are only required if the amp oscillates ultrasonically. Nigel7557's Mapes amplifier oscillated when repaired with epitaxial transistos, but this one may not. Ultrasonic oscillation can be detected with a scope, or with an analog meter. If a measureable AC voltage occurs on output on a VOM while the unit is silent, music will pass through a .47 uf cap but not through a 390 pf cap. Ultrasonic oscillation will blow right through a 390 pf cap. DVM may miss ultrasonic oscillation entirely. The expensive RMS fluke meters are specified to not respond to frequencies over 7000.
 
Last edited:
LE: DC offset hadn't changed after inserting the resistors, but for the idle current the voltage across the test points dropped to about 22mV.
Since I don't know or understand yet how the resistors I placed may change the measurements, I thought any info might help you spot mistakes on my part.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.