That's perfect. I don't have any experience with surface mount devices but that would probably make these really compact if integrated into the design. I'll send you a pm.
I have messed about with some of the footprints of SMD devices but they don't line up neatly with the relatively coarse grid pitch I generally use, so I end up with a layout I'm not happy with. Maybe if I did the whole design using a grid pitch suited to SMD it might come out more pleasing however I have yet to figure it out.
I've never done any circuit layout but if it is like the cad work that I do you can usually change the grid size fairly easily. Don't worry about it I just wondered how much smaller the layout could be with a combination of smd and full size discrete devices in the output section.
Grids can be changed easily, but I like a 0.050" grid for most of my layouts since typical through hole parts have lead spacings in multiples of that pitch. Keeps things tidy. SMD parts on the other hand don't line up nicely at that pitch and what I was suggesting is I haven't found a pitch that works well with a mix of component types.
That is great I have many layout that I want to try out. can't wait to see yours boards 😀
one question man, is possible to order small quantities example 2 or maybe 4 ? 🙂
by the way thank you OS to share your knowledge to us 🙂 I have been checking the layouts really nice work "muy bien" = well done
Regards
Juan
Jason,
Do you have to use the grid or can you place components where you want to? In other words can you do a component layout and then add the traces or do you always use a grid pattern?
Do you have to use the grid or can you place components where you want to? In other words can you do a component layout and then add the traces or do you always use a grid pattern?
You should use grid to place components, unless you forced to do otherwise.
My SMD board is 60x145mm. That's two output pairs and the wolverine IPS, so it does keep the size down a little.
Grid-wise I use a 1mm grid, split into 1/10ths for component and trace snap-to. How small you go depends on how small your parts are. I routinely use 0.5mm pitch parts, where you really do need the extra. If you were going no smaller than 1206 case parts it wouldn't need such a small grid, but the sot23 transistors would probably need smaller. How small would depend on if you wanted to route a single, fine trace, under the transistor.
Grid-wise I use a 1mm grid, split into 1/10ths for component and trace snap-to. How small you go depends on how small your parts are. I routinely use 0.5mm pitch parts, where you really do need the extra. If you were going no smaller than 1206 case parts it wouldn't need such a small grid, but the sot23 transistors would probably need smaller. How small would depend on if you wanted to route a single, fine trace, under the transistor.
You can make the link work for everyone if you remove the page number from the URL. Change this:
Code:
http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble-236.html#post3924790To this:
Code:
http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble.html#post3924790Thanks for the tip!
Now I've got my Wolverine + 2 pair OPS running and everything is okay except for one thing... of course there's always one problem. The zobel likes to get hot when I run highish frequency decent power sine waves through the thing. Say 30kHz+ etc.
I'll start off by saying that I am using 2STC5949s and their complement in the OPS. The drivers are MJE150034/5s and the pre drivers MJE340/350s.
The VAS is a KSC3503 preceded by a 1845.
The VAS CCS is supplied via MJE3x0s.
The IPS is 1845/992s throughout.
I am using R47 RE resistors as I accidentally ordered R02 ohm parts from mouser by mistake 🙂rolleyes🙂 and I only had R47s on hand. R22 are in the post.
Now the good parts, the bias is rock solid and doesn't misbehave in the slightest. Even cranking it up and down to see how the amp reacts doesn't get it into trouble.
There is no evidence whatsoever on the scope of anything oscillating. My scope is only a 50MHz device however, but I know from experience that it can show 'fuzz' much higher than that even if it cannot trigger properly.
THD is excellent all round. We're talking proper blameless performance, with the 2nd and 3rd harmonics at 10kHz rising to around 0.003%, with the higher orders being 0.000x%. Distortion at low frequencies is 0.000x%. These are full power tests into my 9.4ohm dummy load. 10k distortion decreases slightly as the power level drops and by 1 watt is more like 0.001%.
Clipping appears to be excellent.
Apart from the zobel issue it's performing rather well.
Onto the zobel problem. It occurs both if the dummy load is connected or disconnected. Nothing pops up on the scope when this is going on. It is signal level and frequency dependent.
Whenever I am using an amplifier design for the first time I always leave my multimeter connected across an RE resistor so I can monitor the current flowing through the output stage.
Without a load connected this should simply reflect the bias setting, however when I turn on a high frequency sine wave, of decent amplitude, the current flowing through the OPS rises as if it's driving the zobel. If I increase the drive level, the output current also increases in what seems like proportion to the drive level. If I increase the frequency, but keep the drive level the same, the same thing happens, more current flows. Once the input signal is removed the current goes back to the bias setting.
I tried installing the 66pf caps in the optional place on the drivers for stability but this did absolutely nothing.
Quite clearly something isn't quite right, but where to start?
One piece of equipment that I currently lack is a signal generator. I cannot generate square waves, but I can generate sines through the PC.
Now I've got my Wolverine + 2 pair OPS running and everything is okay except for one thing... of course there's always one problem. The zobel likes to get hot when I run highish frequency decent power sine waves through the thing. Say 30kHz+ etc.
I'll start off by saying that I am using 2STC5949s and their complement in the OPS. The drivers are MJE150034/5s and the pre drivers MJE340/350s.
The VAS is a KSC3503 preceded by a 1845.
The VAS CCS is supplied via MJE3x0s.
The IPS is 1845/992s throughout.
I am using R47 RE resistors as I accidentally ordered R02 ohm parts from mouser by mistake 🙂rolleyes🙂 and I only had R47s on hand. R22 are in the post.
Now the good parts, the bias is rock solid and doesn't misbehave in the slightest. Even cranking it up and down to see how the amp reacts doesn't get it into trouble.
There is no evidence whatsoever on the scope of anything oscillating. My scope is only a 50MHz device however, but I know from experience that it can show 'fuzz' much higher than that even if it cannot trigger properly.
THD is excellent all round. We're talking proper blameless performance, with the 2nd and 3rd harmonics at 10kHz rising to around 0.003%, with the higher orders being 0.000x%. Distortion at low frequencies is 0.000x%. These are full power tests into my 9.4ohm dummy load. 10k distortion decreases slightly as the power level drops and by 1 watt is more like 0.001%.
Clipping appears to be excellent.
Apart from the zobel issue it's performing rather well.
Onto the zobel problem. It occurs both if the dummy load is connected or disconnected. Nothing pops up on the scope when this is going on. It is signal level and frequency dependent.
Whenever I am using an amplifier design for the first time I always leave my multimeter connected across an RE resistor so I can monitor the current flowing through the output stage.
Without a load connected this should simply reflect the bias setting, however when I turn on a high frequency sine wave, of decent amplitude, the current flowing through the OPS rises as if it's driving the zobel. If I increase the drive level, the output current also increases in what seems like proportion to the drive level. If I increase the frequency, but keep the drive level the same, the same thing happens, more current flows. Once the input signal is removed the current goes back to the bias setting.
I tried installing the 66pf caps in the optional place on the drivers for stability but this did absolutely nothing.
Quite clearly something isn't quite right, but where to start?
One piece of equipment that I currently lack is a signal generator. I cannot generate square waves, but I can generate sines through the PC.
@ 5th element
Maybe the zobel is just doing it's job !
Here's a Free for personal use SigGen for you that also does Square Waves 🙂 PassMark SoundCheck - Download software to test your PC sound card
Maybe the zobel is just doing it's job !
Here's a Free for personal use SigGen for you that also does Square Waves 🙂 PassMark SoundCheck - Download software to test your PC sound card
The cap of the zobel should have a high impedance at the frequencies we're talking about. At least 40 ohms. The zobel R = 10 ohms, so the resistor would see ≈1/5th the output voltage swing, depending on the frequency. Feeding it 40v p2p would drop 8v p2p or 2.83vrms over the resistor giving 800mW of dissipation. I suppose it is possible.
You can't do proper square waves with a PC as they require high frequency harmonic content that a PC cannot reproduce.
Yes, it's running slightly hot at 5.5mA, this shouldn't affect stability I would not have thought though.
I figured I'd drop this.
8v p-p into the 9.4 ohm load. So what would be 1 watt if this were a standard 8 ohm load.
This is non optimised so to speak as I need to get the lower value RE resistors into the amp and alter the VAS and LTP current a little to get it bang on spec. Still I can't say I am too disappointed 😉
8v p-p into the 9.4 ohm load. So what would be 1 watt if this were a standard 8 ohm load.
This is non optimised so to speak as I need to get the lower value RE resistors into the amp and alter the VAS and LTP current a little to get it bang on spec. Still I can't say I am too disappointed 😉
Attachments
5th element.
2.83V @ 8 ohms is 1 watt output. That would be measured at the output of the amplifier.
2.83V @ 8 ohms is 1 watt output. That would be measured at the output of the amplifier.
Yes read what I said again. 8v p-p. not 8v rms. 8v p-p IS 2.83vrms. = 1 watt into 8 ohm load.
I see this referred to all the time. I have googled it and cannot find it. What exactly is p-p referring to?