Fine 🙂
The lead spacing is both 5 and 7.5 mm and the traditional C9 can accomodate 16mm cap, the alternate one 13mm bu can be stretched to 16mm too.
Fine, let us know your result then 😉
Elna Package
- SoIL4x4 (x1)
- lehmanhill (x1?)
- SoIL4x4 (x1)
Hi,
I was able to do some more testing with the full setup of the old FE beta board Dario was kind of providing me.
As it turned out the sensivity to hum/buzz is mostly electrostatic in nature and that can be handled by proper shielding of the amp inside a metal chassis. The best (and only) connection point should be at the input connectors GND pin (by using a non-isolated RCA jack). The power GND should be left floating (via the 1R it is connected to shield GND). This perfectly feasible with a mono-block approach, and with a stereo amp in one chassis it also no problem when individual mains transformers per channel are used. EDIT: Class-II rated transformer and construction style is required for this and is a good way to build an amp anyway).
Given that the gain of the amp is about 30x and noting that 1mV of hum at the speaker will be audible when listening close, it takes only 30uV (microvolts) of "air-coupled" hum/buzz to get that 1mV. Even when the source resistance is low (I used 50Ohms for the testing) a completely unshielded amp is not optimum.
Magnetic hum can be induced too, when the speaker cables (send and return) are not twisted together and/or run close to the secondary AC lines of the transformer (see my previous post how to perfectly avoid that).
I also did some stability tests, my setup (running from 18Vac == 25Vdc rails) starts to show oscillation when more than 33nF of load capacitance is present. This is a value unlikely to ever be seen with normal speaker cables and lengths, though.
Because of the low supply the clipping Zener protection did not kick in early enough, so clipping the amp caused some awful sounding clicking artifacts (a problem with any nested feedback amplifer topology). I will have to stack two variacs or use different tranny or a lower Zener voltage to check what the clipping protection effectively does (it looks a bit dangerous to me, hard clamping of the LM318's output, but it should be effective nonetheless).So anyone using lower supply voltages might need to use a different Zener (D3) than the stock 27V.
I also did a quick listening test comparing to a standard TDA7294 design, the FE seemed to have quite a bit of more refined midrange/treble (less grain).
I'll try to do some distortion etc measurement the next days. Power supply rejection will be interesting, I expect it to be rather good (compared to conventional chipamps) as it is dominated by the LM318 running off of regulated supplies. That's maybe one of the secrets of the MyRef style amps.
I was able to do some more testing with the full setup of the old FE beta board Dario was kind of providing me.
As it turned out the sensivity to hum/buzz is mostly electrostatic in nature and that can be handled by proper shielding of the amp inside a metal chassis. The best (and only) connection point should be at the input connectors GND pin (by using a non-isolated RCA jack). The power GND should be left floating (via the 1R it is connected to shield GND). This perfectly feasible with a mono-block approach, and with a stereo amp in one chassis it also no problem when individual mains transformers per channel are used. EDIT: Class-II rated transformer and construction style is required for this and is a good way to build an amp anyway).
Given that the gain of the amp is about 30x and noting that 1mV of hum at the speaker will be audible when listening close, it takes only 30uV (microvolts) of "air-coupled" hum/buzz to get that 1mV. Even when the source resistance is low (I used 50Ohms for the testing) a completely unshielded amp is not optimum.
Magnetic hum can be induced too, when the speaker cables (send and return) are not twisted together and/or run close to the secondary AC lines of the transformer (see my previous post how to perfectly avoid that).
I also did some stability tests, my setup (running from 18Vac == 25Vdc rails) starts to show oscillation when more than 33nF of load capacitance is present. This is a value unlikely to ever be seen with normal speaker cables and lengths, though.
Because of the low supply the clipping Zener protection did not kick in early enough, so clipping the amp caused some awful sounding clicking artifacts (a problem with any nested feedback amplifer topology). I will have to stack two variacs or use different tranny or a lower Zener voltage to check what the clipping protection effectively does (it looks a bit dangerous to me, hard clamping of the LM318's output, but it should be effective nonetheless).So anyone using lower supply voltages might need to use a different Zener (D3) than the stock 27V.
I also did a quick listening test comparing to a standard TDA7294 design, the FE seemed to have quite a bit of more refined midrange/treble (less grain).
I'll try to do some distortion etc measurement the next days. Power supply rejection will be interesting, I expect it to be rather good (compared to conventional chipamps) as it is dominated by the LM318 running off of regulated supplies. That's maybe one of the secrets of the MyRef style amps.
Last edited:
While we are waiting for the boards, I would like to suggest an outstanding unit to push the Fremen Edition amps.
Many builders will find simple attenuation like a bare pot or an LDR device to be an excellent approach. For situations that require some gain, I would like to suggest the Jim’s Audio JC-2 kit. It is referenced to the JC-2 designed by John Curl and others and is described as a “straight wire with gain” solution.
I have started a new thread around this package with the intent of getting as many eyes and brains focused on possible component upgrades and optimization.
The preamp sounds fantastic as shipped from the supplier, but a process similar to the ongoing Mini 2948 DAC conversations could possibly lead to even better performance. Please take a look.
http://www.diyaudio.com/forums/analog-line-level/214380-ja-pre-myref-fe-integration.html#post3057400
Many builders will find simple attenuation like a bare pot or an LDR device to be an excellent approach. For situations that require some gain, I would like to suggest the Jim’s Audio JC-2 kit. It is referenced to the JC-2 designed by John Curl and others and is described as a “straight wire with gain” solution.
I have started a new thread around this package with the intent of getting as many eyes and brains focused on possible component upgrades and optimization.
The preamp sounds fantastic as shipped from the supplier, but a process similar to the ongoing Mini 2948 DAC conversations could possibly lead to even better performance. Please take a look.
http://www.diyaudio.com/forums/analog-line-level/214380-ja-pre-myref-fe-integration.html#post3057400
I didn't check the exact level, but it must have been about 23V/31=0.74Vp (0.52Vrms, -6dBV). With most sources (CD) putting out 2Vrms (+6dBV) a preamp will not have to provide any gain at all, so putting a 10k or 50k volume pot ahead will suffice. Note this was with 25V rails, with higher (nominal) rails clipping will start later (and taken care of by the Zener clipping circuit).
I don't think I've ever clipped the original MyRef C at max volume during music playing at max volume from a computer, perhaps I will specifically measure for clipping. But I have halved the voltage loop gain, so maybe that is the reason. My rails are measured to be 27V, but it will vary depending on mains voltage.
If the RCA ground is connected to the chassis, it would also be connected to safety ground, wouldn't it? Think I did something like that initially, and got hum. But I may have connected differently, will give it a try again.
If the RCA ground is connected to the chassis, it would also be connected to safety ground, wouldn't it? Think I did something like that initially, and got hum. But I may have connected differently, will give it a try again.
Last edited:
With smaller amp, preamps etc I never safety ground any chassis but that is only possible because I always use Class-II rated transformers and building techniques (which basically is double insulation so that one faulty isolation cannot cause hazard, only a double failure would do so).
With Class-I xformer and construction of course you MUST earth ground the chassis, then I would use isolated RCA's and attach the earth GND at the power supply GND tab. There are ongoing discussions if partly "isolating" the audio GND from safety GND and chassis via 35A diode bridges is OK (conforming to electrical code) or not. I don't know as I never used these techniques. In tube amps or in really big power amps I always hard ground anything to earth and use input transformers to avoid GND loop problems.
Stability is one of the things that has been neglected in this project in my view. I don't see 33nF as a particularly large load capacitance for a power amplifier.
Unfortunately decent Spice models for the LM318 do not exist and the LM3886 model is limited in scope so computer simulation of gain and phase is not likely to be accurate. Measuring loop gain and phase accurately requires a network analyzer, an expensive piece of equipment.
That leaves monitoring the output voltage of the amplifier with various input frequencies and amplitudes while varying the load impedance. Sometimes low values of load capacitance are more troublesome than high values. Varying the power supply voltage can also turn up an occasional problem. All this is time consuming and offers no guarantee of hitting the right combination of signal and load to show up a performance weakness.
Still, all the reports of the smallest change in components or their placement causing major changes in the sound makes me wonder - is it really stable?
Klaus,
First of all thank you for pointing me in the right direction on the polymer caps, they do indeed leak much more. There are some that get down to the range of 4 times an aluminum electrolytic, but most are very high.
Regarding you post #907, please explain a little more what you mean in grounding a class I transformer. If I understand correctly, you would connect earth ground =? safety ground to the PGND between the power supply caps, C101, C201. Both RCA signal and shield = audio ground would be isolated from the case = safety ground. Audio ground would be connected to the input ground in the upper right of the board (in Dario's pictures) which connects to the ground plane in the upper right corner.
Did I get that right?
Jac
First of all thank you for pointing me in the right direction on the polymer caps, they do indeed leak much more. There are some that get down to the range of 4 times an aluminum electrolytic, but most are very high.
Regarding you post #907, please explain a little more what you mean in grounding a class I transformer. If I understand correctly, you would connect earth ground =? safety ground to the PGND between the power supply caps, C101, C201. Both RCA signal and shield = audio ground would be isolated from the case = safety ground. Audio ground would be connected to the input ground in the upper right of the board (in Dario's pictures) which connects to the ground plane in the upper right corner.
Did I get that right?
Jac
Dario,
Please include me in the cap order
Elna Package (x2)
Ero Package (x2)
Total 18 Euro. Is there any added shipping?
Jac
Please include me in the cap order
Elna Package (x2)
Ero Package (x2)
Total 18 Euro. Is there any added shipping?
Jac
Thanks Klaus for all this measurements and suggestions 🙂
I would have connected to safety earth chassis and tranformers' shield oly, used isolated RCAs and modules left floating (PGND not connected.
Why you would connect RCAs ground to chassis (not safety earthed) and modules' input ground?
The FEs must be used as monoblocks with dedicated transformer for each already (Penasa's suggestion for the double bridge approach).
how much close (speaker cables and secondaries) would induce a significant hum?
The limiter, as I've later discovered, is identical to Penasa's My_Evo one.
The only goal is to avoid a destructive behaviour.
Since the LM318 is supplied with 14V rails shouldn't be the limiter fixed at 28V? 😕
With better parts it's even better 😉
I'm curious to read your listening impressions when you'll have a full stero setup.
It will be an interesting read 🙂
Hi Jac,
no added shipping, those components are light.
Elna Package
- SoIL4x4 (x1)
- lehmanhill (x2)
- SoIL4x4 (x1)
- lehmanhill (x2)
Last edited:
Dario,
You have made favorable mention of the Amtrans caps in the PC section of the BOM in our communications. Could you report more fully your findings, as new builders make their buying choices?
I don't think there is been a lot conversation on those. I placed an order yesterday.
You have made favorable mention of the Amtrans caps in the PC section of the BOM in our communications. Could you report more fully your findings, as new builders make their buying choices?
I don't think there is been a lot conversation on those. I placed an order yesterday.
Thanks again Klaus. You obviously have the most experience in this group.
While you are explaining the grounding to people like me that are a little dense, please comment on grounding the heatsink.
Cheers.
Jac
While you are explaining the grounding to people like me that are a little dense, please comment on grounding the heatsink.
Cheers.
Jac
I was interested in Class II designs, but when I started to search for detachable cords that had 1.25mm2 conductor that could be universally adapted just by swapping the cords, I could find no solution. So I went back to Class I.
I absolutely recommend them for people that wants the ultimate performance, they're the best caps in production among the ones I've tried.
Only NOS KP1834s polystyrenes perform better (more neutral and slightly more refined)
Mouser's BOM FKP2s sound very well but not as refined and neutral as KP1834s
Amtrans are between them (more toward KP1834s): pretty neutral, very refined, natural sounding and musical.
I think the original MyRef design did consider stability, and I believe this variation had not waiver from the original design in this aspect.
The better an amplifier with clean sound will reveal more differences in component variation. This does not relate much with stability.
Just as a general reference, with all the swapping and parts substitution I've done with and to the FEs, I was actually going to mention the ruggedness of this design yesterday. I've made other dumb and dumber slips that weren't reported and am very impressed with the amount of abuse the amps can handle.
I am in no way qualified to engage in the conversations between the technically savvy - you guys are at a deeper level. Just want to share my high opinion of the durability of what has been developed here through the beta build.
I am in no way qualified to engage in the conversations between the technically savvy - you guys are at a deeper level. Just want to share my high opinion of the durability of what has been developed here through the beta build.
You could order parts for a regi-reg, around $5.00 . You power it from one of your PS smoothing caps and it gives a clean 5V supply to power a LDR, a powered Alps pot or even a IR receiver, or all of the above.
Mouser BoM for 5V regulator
R1: 71-RN55D-F-100
R2: 71-RN55D3300F
R3: 71-RN55D-F-499
C1, C3: 80-T356B105K050AT
C2: 80-T350G106K035AT
C4: 667-EEU-FC1V101
D1, D2: 512-1N4002
- Status
- Not open for further replies.