Omicron’s idea is great performance without expensive or unobtanium parts, and without unnecessary complexity.
For example, in this thread a number of people suggested a better - normally more complex and usually more expensive - output stage. In fact, we tested a number of configurations - they all can be made to work - and chose the simplest one that doesn’t compromise overall performance.
Same with opamps - you can use many different opamps, but the good old 5532s do a stellar job at a fraction of the cost of more modern parts.
Oddly, no one suggested a better frequency compensation 😎
Having said that, you can always gild the lily. Omicron works well as is, but you can try better resistors with controlled voltage coefficient for the feedback and crossfeed network; try Nichicon FPCAP or whatever fancies you for power supply rails; roll some opamps (but check for stability and clean clipping into a dummy load as you do). Just make sure to report back here 😉
Why is the offset window large ie - to + 80 mV. The few headphones amps
I build all had offsets of less than a few mV without servos. Is the opamp the culprit or have you designed it for protection of the headphone in case of a major failure?
I build all had offsets of less than a few mV without servos. Is the opamp the culprit or have you designed it for protection of the headphone in case of a major failure?
The width of the comparator window is always a compromise between preventing false triggering at high signal levels and quickly disconnecting the load in case of DC at the output. I would not call the chosen window large, as 80mV @ 32 ohm is 0.2mW, hardly a catastrophic level, and it gets proportionally lower with higher impedance headphones. The actual DC offset with NE5532 is about 1mV.
It all comes from Shakespeare - King John Act 2;
...To gild refined gold, to paint the lily,
To throw a perfume on the violet,
To smooth the ice, or add another hue
Unto the rainbow, or with taper-light [< -- i.e. a candle]
To seek the beauteous eye of heaven to garnish,
Is wasteful and ridiculous excess.
: D
& @alexcp : what a thought-provoking & clearly-articulated project of utmost attention to detail. Thankyou!
How stable is the quiescent current from startup to having reached equilibrium temperature for the heat sinks. I noticed that the diodes d7, d8, d27, d28 are in close proximity of the heatsinks, that I guess would help in stabilizing. If the quiescent current proves relatively stable than I assume there is no need for a control configuration of it like in Passes Whammy and others.
Indeed, the diodes are located on the PCB close to the heatsinks, and that provides good bias stability. Some builders prefer to place these diodes above the PCB with longer leads and make them touch the heatsink or the transistor, but I didn’t find that necessary.
Hi Alex,
looking over the BOM for the SMT version, I noticed that there is no space for the RLRC filter network on the SMT PCB? What is the reason for this? Is it not required or should we add it free hanging / on the 6.35 mm jack?
looking over the BOM for the SMT version, I noticed that there is no space for the RLRC filter network on the SMT PCB? What is the reason for this? Is it not required or should we add it free hanging / on the 6.35 mm jack?
Thanks. Somehow I missed that post and the left part in the schematic in picture 3 in post #167
I spotted an error in the BOM for the SMT board - I am sorry about this.
The BOM (post #172) says that C13, C14, C33, C34 can be up to 13mm diameter and gives an example part number that is 12.5mm, which is incorrect - the smaller SMT board was designed for the large caps 10mm in diameter, and an appropriate part number from the Panasonic FC series is the taller (30mm tall) EEU-FC1E102L. Should you not want such a tall part on the board, you can choose e.g. 680uF part such as EEU-FC1E681. Both parts are currently available at Mouser, and other caps with 10mm diameter can be used.
If you already own or ordered 13mm caps, you may still be able to fit them, as the lead spacing is the same (5mm), but perhaps they would need to be installed a couple of mm above the board to allow for nearby surface mounted parts - I will be able to verify this early next week.
I apologize for the inconvenience.
The BOM (post #172) says that C13, C14, C33, C34 can be up to 13mm diameter and gives an example part number that is 12.5mm, which is incorrect - the smaller SMT board was designed for the large caps 10mm in diameter, and an appropriate part number from the Panasonic FC series is the taller (30mm tall) EEU-FC1E102L. Should you not want such a tall part on the board, you can choose e.g. 680uF part such as EEU-FC1E681. Both parts are currently available at Mouser, and other caps with 10mm diameter can be used.
If you already own or ordered 13mm caps, you may still be able to fit them, as the lead spacing is the same (5mm), but perhaps they would need to be installed a couple of mm above the board to allow for nearby surface mounted parts - I will be able to verify this early next week.
I apologize for the inconvenience.
Getting ready to place my order, I noticed the BOM for the PSU lists L1 as 11mH SU9VD, but all the SU9VD parts I find are in the uH ranges. PSU schematic shows SS11 as the part number and there is a 12mH SS11 Kemet inductor. Can you clarify which I should be ordering here?
Thanks!
Thanks!
It appears that the “D” in SU9VD denotes”high frequency type” with a different core type and a lower inductance. Higher inductance chokes from that series do not have the ”D”, i.e. the 10mH SU9V-01100. Sorry about this confusion.
Note that KEMET SU9 series is not the only option, e.g. I built one board with Wurth 7448640418 choke, and there are lots of other suitable parts.
Note that KEMET SU9 series is not the only option, e.g. I built one board with Wurth 7448640418 choke, and there are lots of other suitable parts.
Yes, 13mm caps will just fit it mounted on top of nearby SMT parts:
,
Still, 10mm caps are recommended.
Sure thing. The code after the comma refers to how this transistors are packed. The transistors themselves are identical.