The KEF Q350 is a bookshelf entry level 2 way ported speaker featuring the KEF SP 1709 coaxial driver. If often goes on sale and was now replaced with the very similar KEF Q 3 model adding KEF’s meta technology. These are my DIY measurements and tweaks for the Q350. You may find it interesting
Port tuning
To me the most critical part of the Q350 is to get the bass response right in your room. If you need to place it close to a wall or corner and it sounds too boomy with the port open or half open but lacking deep bass with the port closed you can experiment with extending the port length. I used a 19cm peace of slotted pipe isolation from the hardware store (outer diameter ~6cm, inner diameter ~4.2cm) fixed with some duct tape and plugged it into the port about 5cm deep (which reduced the inner diameter a bit). To fine tune it I added a little bit polyester wool on the outlet of the tube and fixed with a rubber band.
Damping
There is little damping material in the Q350. Just 2 rolled up sheets of polyester wool. The rolls stick between the sidewalls. One role is located at the bottom and the other one located at the top (glued to the sidewall to not fall down and cover the port).
The Q350 sounds a little bit under damped, hollow, resonant, boomy or warm to me, so I added 2 sheets of polyester wool to cover the rear panel and added one more roll of polyester wool to the bottom.
Cabinet
The cabinet of the Q350 is really basic. It is made of thin 16mm MDF, with a 35mm sandwich front baffle with simple plastic veneer. There is one bracing in the middle running from left to right sidewall. Front and rear baffle are each connected with 4 triangular bars to both sidewalls. 2 more triangular bars connect the bottom to one sidewall.
To strengthen this rather lightweight construction I added 10 more triangular bars (47mm long cut from 18mm MDF):
Crossover
The crossover of the Q350 is also very basic, a first order filter for woofer and tweeter and a single resistor to adjust the tweeter level, with entry level quality parts (0,957mH 0,4 Ohm ferrite coil inductor, 4,444uF foil conductor, 1,501 Ohm ceramic resistor). Anyway it achieves a pretty good summation of the drivers and at least it avoids electrolytic capacitors. But when looking at the individual driver response I noticed that the woofer plays up to 10 khz rather loud with peaks in the 7-10 khz region which the 1 order network cannot fully reduce so they show up a little bit even in the driver summation. So I decided to design a steeper 2 order rolloff for the woofer but with very similar summation result instead of just replicating the original crossover with higher quality parts. To me the audible difference to the original crossover in a short Mono A/B listening session is rather subtle.
The next picture shows the material (one speaker) for the above tweaks and the original crossover:
For the measurement I used affordable DIY gear, a Focusrite 2i2 3 gen sound interface , self calibrated Behringer ECM8000 microphone, self build rotary table (https://www.amazon.de/dp/B0BLHHF15Z, https://www.amazon.de/dp/B0BYYN2ZM4), REW, VituixCAD and followed REW VituixCAD guide.
To check how accurate this measurement and simulation really is lets compare Amir’s on axis measurement with my complete speaker and individual driver + VituixCAD crossover simulation:
The blue line on top is Amir’s result, in the middle the measured result of the complete speaker with original (green) and new crossover (red). On the bottom you see the result of the individual driver measurement with VituixCAD crossover simulation. I am very happy with this result, it matches reasonably well with the professional measurment and the VituixCAD crossover simulation is very accurate.
Lets have a look at the on axis driver summation simulated in VituixCAD for some crossover designs:
Feel free to download my VituixCAD data to investigate more. If you want to do your own measurement of individual drivers for VituixCAD crossover simulation I strongly recommend to carefully read the instruction.
I would suggest to test your measurement setup for correctly time aligned measurements once before executing all these many measurements. To do so, just perform an on axis 1m far field measurements of the entire speaker and all it’s individual drivers (with the crossover still in place) and apply a 4 ms time window to all measurements. Use REW trace arithmetic to sum all individual drivers measurements. If your setup is correct the summation should match the complete speaker measurement perfectly, if not you will see small or huge derivations especially in the crossover regions.
- To see how good they match with Amir’s professional Klippel measurements
- To see how good VituixCAD crossover simulation works
- Try out my affordable port tuning, damping, cabinet bracing or crossover tweaks
- Use my VituixCAD data to start you own Q350 passive or even active crossover designs
Port tuning
To me the most critical part of the Q350 is to get the bass response right in your room. If you need to place it close to a wall or corner and it sounds too boomy with the port open or half open but lacking deep bass with the port closed you can experiment with extending the port length. I used a 19cm peace of slotted pipe isolation from the hardware store (outer diameter ~6cm, inner diameter ~4.2cm) fixed with some duct tape and plugged it into the port about 5cm deep (which reduced the inner diameter a bit). To fine tune it I added a little bit polyester wool on the outlet of the tube and fixed with a rubber band.
Damping
There is little damping material in the Q350. Just 2 rolled up sheets of polyester wool. The rolls stick between the sidewalls. One role is located at the bottom and the other one located at the top (glued to the sidewall to not fall down and cover the port).
The Q350 sounds a little bit under damped, hollow, resonant, boomy or warm to me, so I added 2 sheets of polyester wool to cover the rear panel and added one more roll of polyester wool to the bottom.
Cabinet
The cabinet of the Q350 is really basic. It is made of thin 16mm MDF, with a 35mm sandwich front baffle with simple plastic veneer. There is one bracing in the middle running from left to right sidewall. Front and rear baffle are each connected with 4 triangular bars to both sidewalls. 2 more triangular bars connect the bottom to one sidewall.
To strengthen this rather lightweight construction I added 10 more triangular bars (47mm long cut from 18mm MDF):
- 1 Bottom to Front
- 1 Bottom to Rear
- 2 Bottom to other sidewall
- 1 Top to Front
- 1 Top to Rear
- 4 Top to both sidewalls
Crossover
The crossover of the Q350 is also very basic, a first order filter for woofer and tweeter and a single resistor to adjust the tweeter level, with entry level quality parts (0,957mH 0,4 Ohm ferrite coil inductor, 4,444uF foil conductor, 1,501 Ohm ceramic resistor). Anyway it achieves a pretty good summation of the drivers and at least it avoids electrolytic capacitors. But when looking at the individual driver response I noticed that the woofer plays up to 10 khz rather loud with peaks in the 7-10 khz region which the 1 order network cannot fully reduce so they show up a little bit even in the driver summation. So I decided to design a steeper 2 order rolloff for the woofer but with very similar summation result instead of just replicating the original crossover with higher quality parts. To me the audible difference to the original crossover in a short Mono A/B listening session is rather subtle.
The next picture shows the material (one speaker) for the above tweaks and the original crossover:
For the measurement I used affordable DIY gear, a Focusrite 2i2 3 gen sound interface , self calibrated Behringer ECM8000 microphone, self build rotary table (https://www.amazon.de/dp/B0BLHHF15Z, https://www.amazon.de/dp/B0BYYN2ZM4), REW, VituixCAD and followed REW VituixCAD guide.
To check how accurate this measurement and simulation really is lets compare Amir’s on axis measurement with my complete speaker and individual driver + VituixCAD crossover simulation:
The blue line on top is Amir’s result, in the middle the measured result of the complete speaker with original (green) and new crossover (red). On the bottom you see the result of the individual driver measurement with VituixCAD crossover simulation. I am very happy with this result, it matches reasonably well with the professional measurment and the VituixCAD crossover simulation is very accurate.
Lets have a look at the on axis driver summation simulated in VituixCAD for some crossover designs:
Feel free to download my VituixCAD data to investigate more. If you want to do your own measurement of individual drivers for VituixCAD crossover simulation I strongly recommend to carefully read the instruction.
I would suggest to test your measurement setup for correctly time aligned measurements once before executing all these many measurements. To do so, just perform an on axis 1m far field measurements of the entire speaker and all it’s individual drivers (with the crossover still in place) and apply a 4 ms time window to all measurements. Use REW trace arithmetic to sum all individual drivers measurements. If your setup is correct the summation should match the complete speaker measurement perfectly, if not you will see small or huge derivations especially in the crossover regions.
Last edited:
