OK, no more questions.
Can't wait building one. I'll definitely use 1x as low gain setting because I have sensitive headphones, sensitive ears and never understood the craze for more power than is ever going to be needed. What's the point. It's just a waste of volume control range and, with designs where it is in front of the gain stage, adding noise, channel imbalance etc. x_x
gain and graphs
Here is the gain issue graphically.
The LT Spice library doesn't have the njm2068 of course. I've substituted in the LT1469-2 which is one of Groner's suggested upgrades for an NE5532. The LT1469-2 is good for up to +/-18v also, so it works just fine to show the clipping issue at 12v vs. gain and input signal level.
All plots are with +/-12V rails, assuming the "home" ac-powered case. The circuit is just for one channel since both are identical, of course. The legend for the plots are at the top of each and the file names (hover the mouse over each) shows the specifics.
The first set of plots are with the original 3.1x gain setting with the 715R resistor. The first plot is with Vin = 2Vrms = 2.82Vpeak input at 2Khz. No clipping (the output is red). The second plot then is for Vin = 2.7Vrms = 3.82Vpeak input at 2Khz. The clipping on the output is apparent.
The second set of plots are with the new 2.5x gain setting with a 1k resistor. The first plot is with Vin = 2Vrms = 2.82Vpeak input at 2Khz. No clipping. The second plot then is for Vin = 2.7Vrms = 3.82Vpeak input at 2Khz. No clipping this time.
The final two plots are the new 2.5x gain with 2Vrms in at 10hz and 30Khz.
added 2nd stage
...and here is an approximation of the O2 second stage tacked on to take a look at clipping behavior end to end. I've had to use Linear Technology substitutes again, due to lack of njr models, from the LT Spice library.
The LT1469 (single, without the -2) is used since it is unity gain stable (-2 needs minimum gain of 2), driving a LT1010 150mA current buffer. The single LT1469 + LT1010 150mA pair "replaces" the two parallel njm4556 in one channel for simulation purposes, which are 70mA x 2 = 140mA.
The volume pot is on "full" in both plots for maximum signal transfer. The load is a resistive approximation of my 44R pair of Shure SRH-940s. :D
The first plot is of the original 3.1x gain is used with the 2.7Vrms input to plot clipping behavior end to end. The red is now the output of the second stage and blue is the output of the first stage.
The second plot is with the new 2.5x gain and a 2Vrms signal, showing no clipping end to end. Notice here the red output plot is sitting pretty much right on top of the first stage plot, a very good thing in current-buffer land. :) Note on the second plot that 7V peak is 5Vrms, which would be 5^2 / 44 = 550mW per channel, something my ears would never forget. :) Plenty of output at 44R with 2Vrms in and the first stage gain of 2.5x.
Lol - won't likely have a chance to log in much if any next week so probably won't see any posts/replies until the weekend.
First, thanks agdr for the sims! I based my numbers off real measurements. On AC power it's pretty easy. On battery power, however, the clipping point is battery voltage dependent. But we seem to be pretty close to being in agreement.
NEW FILES! - Revised files can be found in the O2 Files & Resources section and there are also now PCB files!
IMPORTANT - For anyone wanting to rush out and get boards made, you might want to hold off at least a few days. Extra sets of eyes might catch something I missed. Also, I should have more information soon about the availability of the board from one or more commercial sources. I know some have also talked about a group buy.
DESKTOP AMP - I believe the trade offs I made with the O2 in the interest of cost, size, and portability are worthwhile. I'm not aware of another portable amp that can do a better job overall--let alone for under $40 - $100. But...
There are things that can be improved in a desktop version. One of those is changing the gain structure to eliminate the input overload issue. In the O2 there simply isn't room to solve the problem without significantly hurting the performance.
Eliminating the batteries frees up lots of board space and removes $11+ from the total cost. That space, and the $11, can be put to good use improving the audio performance still further.
CHOICES - If you intend to use the O2 only where there's AC power, and you're not in a hurry, you might want to wait for the Desktop version.
COST - I don't know where the total cost will come in for an upgraded desktop version. The biggest issue will be the front and rear panels if you don't want to drill your own. Needing two panels instead of one will add around $20 to the total cost. I would expect the complete desktop amp to still be well under $130 and, if you get clever with the enclosure, you could probably build one for under $70.
TIMING - I also don't know when I'll get the desktop design published. At least I don't need to write another 40,000 words documenting it! I'll mostly just note where it differs from the O2. Still, there will likely be a couple spins of the PCB, prototypes built, lots of measurements have to be made, and there still will be a lot to document and write up. With the O2 cat already out of the bag, the desktop version is likely to be a more collaborative effort.
MORE DETAILS - I'll be sharing more details about the desktop version soon to see what everyone thinks. I have a bit more work to do first to know what all will be possible keeping some basic constraints in place.
I really wish you would reconsider on the NiMh batteries though. LiFePO4 are really much better and just as cheap. Charging conditions are slightly more stringent. Two voltage regulators and a few resistors are all one needs to setup a voltage and current controlled charging solution similar in function to the one you have there.
I realize that is quite a crude example. It is feasible though. 6s center tapped is just the ticket. It is less likely to become imbalanced because it is fewer cells in series. The 9v NiMh are 8 in series and you are adding another 8 for dual PS. 6 vs 16 is a better proposition.
All of this nonsense about batteries burning up has got to end sometime. Have you ever seen A NiMh cell with an internal short? It's fireworks much the same.
Please show me where I can get SIX high quality LiFePO4 cells (not the exploding no-name Chinese rejects on eBay) for under $11 (the cost of two NiMh 8.4V batteries) including shipping costs (you said "just as cheap").
Actually it's quite a bit less than $11 because you have to subtract all the extra charging/safety circuitry.
Then you have to find a way to mount and secure the six cells. And is there room for them and the extra charging circuitry in the existing enclosure?
The you have to redesign the entire PC board, that's already had a zillion hours invested in 3 revisions, to accommodate the new form factor of the batteries and the new charging circuitry.
Then factor in the DIY-unfriendliness and higher safety risks. I can short out a 9 volt Ni-Mh battery and it just gets hot and dies as the entire battery fully charged only has about 1.5 Watt/hours of total energy. LiFePO4 batteries are not nearly as polite.
And will the LiFePO4 amp work any better? It might have longer run time but there's already a 25+ hour version of the O2 for those who prioritize run time above all else. But will it measure better? I doubt it.
Let's get a group buy rolling!!
I do understand what you are saying about having to redo a whole crap ton of work. I know that isn't going to be a possibility for this generation of the O2 but it is something to think about.
I checked it out, and 3s LiFePO4 already packaged with protection PCB that protects against direct short/over/undercharge is $12.
LiFePO4 18500 Battery: 9.6V 800 mAh (3x18500 Flat, 7.68Wh, 7.0A rate ) with PCB & Polyswitch
So that's a $24 initial investment for a battery with more than two times the run time. It has also got close to a 3v edge over the NiMh. By the time you have replaced the NiMh batteries two times, the LiFePO4 would still have plenty of life left. Close to 500 more charges. That is over a year of everyday use. I bet it is actually more. That figure is low balled quite a bit.
$11 battery now, that will cost me 11$ to replace every year and gives half the run time at a lower voltage VS a $24 battery that I would need to replace every three years. Not looking good on price. Looks to me like the LiFePO4 is somewhere around 2/3 as expensive over time. Lets not even talk about charging time and internal resistance.
You have already got the LVC working. It could easily be changed to accommodate. The only thing missing is the power supply/charger. It won't be as easy as the clever half wave supply you have setup for the O2. It's not really that much harder either. Like I said, it's really my only critique of the design. It's not going to stop me from building one, but the battery is really the most important part of a portable amp.
|All times are GMT. The time now is 10:42 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio