Hi Mooly
Thanks again -- Looking through my FET stocks, I find that I now have 2 pairs of the 2SK135/2SJ50 Devices. These use the Exact Same Datasheet as the Original (missing) 2SK133/2SJ48 Devices, except that of course they are Higher Voltage, at 160 Volts as against the original 120.
Do You think before I start to add those capacitances as discussed, that I should perhaps replace those 2SK176/2SJ56 Output Devices with the 2SK135/2SJ50 ones?
It's easy, as they are only Socketed In, no Desoldering --
Thanks Again --
Thanks again -- Looking through my FET stocks, I find that I now have 2 pairs of the 2SK135/2SJ50 Devices. These use the Exact Same Datasheet as the Original (missing) 2SK133/2SJ48 Devices, except that of course they are Higher Voltage, at 160 Volts as against the original 120.
Do You think before I start to add those capacitances as discussed, that I should perhaps replace those 2SK176/2SJ56 Output Devices with the 2SK135/2SJ50 ones?
It's easy, as they are only Socketed In, no Desoldering --
Thanks Again --
Hello, All --
Well regarding the Hitachi hA6800 Amp Problem, swapped out the Output Fets for the older version, No Change to the situation. While looking more closely at the Motherboards, I noticed that 3 of the 2SC2389 transistors looked different and perhaps newer than the others, maybe as a result of work done here before me. Thinking that this could possibly be contributing to the Oscillation Fault, I ordered and then replaced them all. Then; Big Change. Still Oscillation, but Far less in amplitude -- Now, it was Only about 2 volts peak to Peak, but, still on Both Channels. On Studying the layout of the motherboard, I saw that there is a Marked Location for a C707, (not on the drawing); connecting the Collector of Q707 to the Base of Q705, (in the same manner of C704 between the C of Q706 and the B of Q704). This gets us into the Exact Area of the Hitachi HA 4700 who's Oscillation problems were discussed earlier on this post.
So, I decided to add a 56pf Capacitor into that location in both Channels. Powered up. Lo and behold, still Oscillation but much less in amplitude, now down to 400MV P-P on both channels. This Oscillation is going off at about 10.2 MHz, according to my F/counter. When You inject a 1 Khz Sinewave into the input, this is produced properly, but with the Oscillation riding on the 1 Khz Output Waveform. I then tried Doubling the Value of those added capacitors, but this only made the amplitude of the Oscillation higher.
While I was at it, I also tried adding Capacitors of 100 Pf between the Gates and Drains of all the Output FETs, which also made things worse. The Idle Current sets up perfectly, and is stable, so no problem in that area. -- So, Where to from here?
Are there any other things that I can do to help the situation?
Any and All Help would be greatly Appreciated.
Attached: HA6800 Schematic
Well regarding the Hitachi hA6800 Amp Problem, swapped out the Output Fets for the older version, No Change to the situation. While looking more closely at the Motherboards, I noticed that 3 of the 2SC2389 transistors looked different and perhaps newer than the others, maybe as a result of work done here before me. Thinking that this could possibly be contributing to the Oscillation Fault, I ordered and then replaced them all. Then; Big Change. Still Oscillation, but Far less in amplitude -- Now, it was Only about 2 volts peak to Peak, but, still on Both Channels. On Studying the layout of the motherboard, I saw that there is a Marked Location for a C707, (not on the drawing); connecting the Collector of Q707 to the Base of Q705, (in the same manner of C704 between the C of Q706 and the B of Q704). This gets us into the Exact Area of the Hitachi HA 4700 who's Oscillation problems were discussed earlier on this post.
So, I decided to add a 56pf Capacitor into that location in both Channels. Powered up. Lo and behold, still Oscillation but much less in amplitude, now down to 400MV P-P on both channels. This Oscillation is going off at about 10.2 MHz, according to my F/counter. When You inject a 1 Khz Sinewave into the input, this is produced properly, but with the Oscillation riding on the 1 Khz Output Waveform. I then tried Doubling the Value of those added capacitors, but this only made the amplitude of the Oscillation higher.
While I was at it, I also tried adding Capacitors of 100 Pf between the Gates and Drains of all the Output FETs, which also made things worse. The Idle Current sets up perfectly, and is stable, so no problem in that area. -- So, Where to from here?
Are there any other things that I can do to help the situation?
Any and All Help would be greatly Appreciated.
Attached: HA6800 Schematic
This amplifier is over 30 years old and the Output Devices are housed in plug-in sockets a not uncommon practice back then that has fallen out of favour. I doubt the contacts are as pristine as they were when new - quite possibly a long way apart. In my view electrically and mechnically these parts are not to be relied upon.
Try soldering the Output Device connections directly to the pcb.
Gate stopper resistors work best when these are close to the gate pin - a move suggested earlier in post 24.
You could try this, you could increase the stopper resistor value a little if you have to.
I'm afraid I can't add much beyond my thoughts in post #40. The design must have worked when using its original parts
No magic answers I'm afraidAdding capacitance at strategic points in a circuit will often tame instability, but ultimately the reasons why this has to be done really need addressing.
For example, whether it really is a marginal design stability wise, or whether alternate semiconductor replacements have changed the characteristics and altered the phase margins.
Thanks for the suggestion re those Mosfet Device Sockets; I pulled them and serviced them, all looks okay, still did not help. Then, I decided to do some reading of the T/I publication AN1645, as suggested in Post #33, and act on some of it's findings.
I tried changing the Values of the Gate Stopper Resistors as suggested in AN1645.
I Changed R720 to 1500 Ohms, and R721 to 1000 Ohms. Problem Gone, utterly --
Then, I tried reducing thse Values to see what the effect was. I ended up by using R720 as 1000 Ohms, and R721 as 680. I then tried removing the C707 which I had fitted earlier, but was again met with Oscillation in Both Channels. I then refitted those caps on the top side of the Pcb, and now the result is absolutely stable operation.
I have reset the IQs to the correct value, (as they did require a little adjustment when there was no oscillation present), and let her warm up properly to a stable temperature.
So far, there is no sign of Any Oscillation on my 'scope's Most Sensitive setting.
Tomorrow, I intend to do the Industry Standard Power Output Test of 1 Khz into 8 Ohms to see what happens. After that, I will be doing an Input Frequency Sweep to see what the results are. I will be posting the results. I don't know why I have had to change those Gate Stopper Values, but it has certainly done the trick. The only thing I can think of is that the Output Devices I have used are still not what the Amp was built around, which of course were 2SK133 and 2SJ48. I am using 2SK135 and 2SJ50.
Thanks so much for all Advice and Help Given, wish I can be of help to someone in the future!
I tried changing the Values of the Gate Stopper Resistors as suggested in AN1645.
I Changed R720 to 1500 Ohms, and R721 to 1000 Ohms. Problem Gone, utterly --
Then, I tried reducing thse Values to see what the effect was. I ended up by using R720 as 1000 Ohms, and R721 as 680. I then tried removing the C707 which I had fitted earlier, but was again met with Oscillation in Both Channels. I then refitted those caps on the top side of the Pcb, and now the result is absolutely stable operation.
I have reset the IQs to the correct value, (as they did require a little adjustment when there was no oscillation present), and let her warm up properly to a stable temperature.
So far, there is no sign of Any Oscillation on my 'scope's Most Sensitive setting.
Tomorrow, I intend to do the Industry Standard Power Output Test of 1 Khz into 8 Ohms to see what happens. After that, I will be doing an Input Frequency Sweep to see what the results are. I will be posting the results. I don't know why I have had to change those Gate Stopper Values, but it has certainly done the trick. The only thing I can think of is that the Output Devices I have used are still not what the Amp was built around, which of course were 2SK133 and 2SJ48. I am using 2SK135 and 2SJ50.
Thanks so much for all Advice and Help Given, wish I can be of help to someone in the future!
So this result adds weight to the differing characteristics of the semiconductors being all important.
The circuit right at the beginning of this thread seems to show that a normal Zobel network isn't fitted. It might be worth looking into that.
When testing the amp I would also look at how it behaves when squarewave tested as that can often provoke instability or conversely show whether the design is basically stable.
The circuit right at the beginning of this thread seems to show that a normal Zobel network isn't fitted. It might be worth looking into that.
When testing the amp I would also look at how it behaves when squarewave tested as that can often provoke instability or conversely show whether the design is basically stable.
good point about the missing zobel, Mooly.
I think the original schematic from Hitachi's app note in HLN600 as well the write-up in IEEE Transactions on Consumer Electronics (?) when the lateral MOSFETs were introduced years ago both mentioned that the output zobel was required.
mlloyd1
I think the original schematic from Hitachi's app note in HLN600 as well the write-up in IEEE Transactions on Consumer Electronics (?) when the lateral MOSFETs were introduced years ago both mentioned that the output zobel was required.
mlloyd1
Last edited:
wow, I found the article i referred to.
It's very old - i had just started college.
google truly is amazing:
100 Watt Super Audio Amplifier Using New Mos Devices - IEEE Journals & Magazine
mlloyd1
It's very old - i had just started college.
google truly is amazing:
100 Watt Super Audio Amplifier Using New Mos Devices - IEEE Journals & Magazine
mlloyd1
The gate to source capacitance in combination with the gate stopper has a low pass filtering effect. There could be some difference in the former characteristic between the MOSFET types however the stopper resistors could be equally to blame.
These are fusible types (also flameproof) which may have been degraded by heat over the life of the amplifier and although they are metal film the construction could have an element of inductance at VHF that disappears with a normal metal film replacement - helix cut to shorter length possibly.
Last edited:
Hello
Thanks for that -- If You look at the Schematic in Post #44, You can see the Zobel, or Snubbing Network. It's L702, .69uH, in parallel with R728, 10 Ohms to the Output Terminal, which is then snubbed to ground by R730, 4.7 Ohms, in Series with C713, .047uF.
Is this what You are referring to? --
Thanks for that -- If You look at the Schematic in Post #44, You can see the Zobel, or Snubbing Network. It's L702, .69uH, in parallel with R728, 10 Ohms to the Output Terminal, which is then snubbed to ground by R730, 4.7 Ohms, in Series with C713, .047uF.
Is this what You are referring to? --
If this question is directed at me the answer is no however now you have raised this oscillation at VHF can cause resistor and capacitors in a Zobel network to fail - a mains rated capacitor would be better than a 100 V mylar.
I looked at the schematic in the area of the stopper resistors. The values you have substituted are multiples of the original values - R720 is 100R with 47pF in parallel while R721 is 10R.
If you increase the values of these too much you will reduce the frequency limit and phase margin will reduce. It is not clear what the values you started with but you have moved well away from the schematic.
You have a Miller cap C704 for the dominant pole and lead compensation of 5pF, a series composite of C722 and C714, to fine tune the phase.
At VHF where phase may be in excess of 180 degrees the latter series pair of caps have a low impedance connection to inverting input terminal.
If increasing the stopper values is a necessary approach you might reach a point where C722 and C714 become unnecessary.
Mooly has suggested running a square wave test into a test load. You could give that a shot at 1kHz.
Hello
Thanks for that. Do You think that it's possible that because I have substituted those Output Devices that this change has become necessary? I had previously built other Mosfet Amps from scratch and had similar problems with these same Mosfet Devices, which I had overcome in the same way.
Also, At the time, I was unable to find those proper Fusible Resistors you spoke of, and indeed the Amp now just has 1/8 watt Carbon Resistors in those locations, Do You think it would be worth swapping these out for 1/8 watt Metal Film Devices?
Last Tuesday, I Started off by substituting 470 ohms for R720 and 100 for R721, no changeto the Oscillation. And then decided to go to 1500 Ohms, and 1000 Ohms, (as above), which then cured it completely.
I then reduced them to 1000 ohms and 680, values I saw in another Commercial Amp. Another thing you may not have seen is that Capacitor C707, which is marked on the Pcb, but not present in the Hitachi Circuit, I had added as 56 pF, and is still in there on Both Channels. -- If I then remove it, Oscillation will return, as below --
But the original Oscillation was 'only' 400 mV on both channels before I started --
Referencing a Square Wave Test; would You suggest doing this at Full Power, or just at a lower level first to see how it looks?
Obviously You have a very good insight into how these things should operate in principle, something I should perhaps study up some more so as to be aware --
Thanks so much for your Insight/help!
Thanks for that. Do You think that it's possible that because I have substituted those Output Devices that this change has become necessary? I had previously built other Mosfet Amps from scratch and had similar problems with these same Mosfet Devices, which I had overcome in the same way.
Also, At the time, I was unable to find those proper Fusible Resistors you spoke of, and indeed the Amp now just has 1/8 watt Carbon Resistors in those locations, Do You think it would be worth swapping these out for 1/8 watt Metal Film Devices?
Last Tuesday, I Started off by substituting 470 ohms for R720 and 100 for R721, no changeto the Oscillation. And then decided to go to 1500 Ohms, and 1000 Ohms, (as above), which then cured it completely.
I then reduced them to 1000 ohms and 680, values I saw in another Commercial Amp. Another thing you may not have seen is that Capacitor C707, which is marked on the Pcb, but not present in the Hitachi Circuit, I had added as 56 pF, and is still in there on Both Channels. -- If I then remove it, Oscillation will return, as below --
But the original Oscillation was 'only' 400 mV on both channels before I started --
Referencing a Square Wave Test; would You suggest doing this at Full Power, or just at a lower level first to see how it looks?
Obviously You have a very good insight into how these things should operate in principle, something I should perhaps study up some more so as to be aware --
Thanks so much for your Insight/help!
The network on the end of L702 isn't functioning as a Zobel, more correctly its more like a Thiele network.
A Zobel network is a series R/C network before the coil. I don't know why this design hasn't got that because its something that most amplifiers need for stability.
Squarewave testing is very tough on an amp and should be done at low power. Things get hot very quickly once you get over a few watts.
A Zobel network is a series R/C network before the coil. I don't know why this design hasn't got that because its something that most amplifiers need for stability.
Squarewave testing is very tough on an amp and should be done at low power. Things get hot very quickly once you get over a few watts.
Thiele networks have been around for decades and used in popular DIY power amplifier projects in Australian electronics magazines and sold in kit set form.
These networks are supposed to guarantee unconditional stability. That raises a question if the trouble arises within the Hitachi electronics or the external stability network.
In the last regard there seems to have been an assumption that the lower resistance in the coil makes R728 immune to damage and therefore the resistance need not be measured.
Carbon film resistors can go high in value under stress or burn out without showing visible signs. R728 in parallel with the output coil is a 0.25W Carbon film. It is included to reduce the Q of the coil and damp ringing effects. From the evidence one cannot be sure of that.
Silicon Chip designers don't take chances and use higher rated components than those in the Hitachi circuit.
My advice is to Telnet 100 is to lift one end of R728 and measure the resistance.
Hello, All--
Thanks for the helpful advice. I was always under the belief that any network connected to the output terminals of an Amp was called a Zobel Network -- Live and learn!
I did as suggested and lifted one end of both of those R728 in both channels; both measure as they should at 10 Ohms. Further to this, I saw that the PC Board is both Marked and Tracked for an R729, in series with a C712, to ground. This is (as Mooly suggested) connected BEFORE the Coil L702. These Components are not shown on the Hitachi Circuit, so Someone at Hitachi research must have been going to use this circuit. In the interests of further stability, would it be a good idea to add these components, making the Values 10 Ohms 1 watt, and .1 uF, 250v? -- I await your well informed advice --
Thanks for the helpful advice. I was always under the belief that any network connected to the output terminals of an Amp was called a Zobel Network -- Live and learn!
I did as suggested and lifted one end of both of those R728 in both channels; both measure as they should at 10 Ohms. Further to this, I saw that the PC Board is both Marked and Tracked for an R729, in series with a C712, to ground. This is (as Mooly suggested) connected BEFORE the Coil L702. These Components are not shown on the Hitachi Circuit, so Someone at Hitachi research must have been going to use this circuit. In the interests of further stability, would it be a good idea to add these components, making the Values 10 Ohms 1 watt, and .1 uF, 250v? -- I await your well informed advice --
R729 and C712 are not in the parts list. It might have been intended to use these instead of the Thiele network components but the value of L1 would need to be changed - I don't see 0.69uH as being enough.
Now that you have restored R728 in position there is still one hurdle to clear with the output stability network - this is to run a sine wave signal into a test load and use your 'scope to measure the a.c. voltage drop across R730.
You need to be sure that C713 is working properly under a.c. conditions.
Looking at schematic the source resistors R722 and R723 have values of 0R22 which would be correct. On the other hand the parts list shows the value of R722 as 10R. One would hardly believe that is the value of the actual part on the board. R723 is correctly listed. This does raise the question if there are other anomalies or not.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.