This article documents improvements to the Manson EP-613 / Jaytech MP-3082 power supply, improvements motivated by failure of one of the supplies due to excessive temperature rise on a load that was within specifications. Yet another Manson design failure.
The MP-3082 components damaged in the heat induced failure were:
On replacing Q1 is was noted that the screws securing the transistor to the heatsink were not very tight, and on checking, neither were the other transistor's screws. Neither transistor was properly mounted, and this probably contributed to the failure.
On repairing the power supply, it was put to test on a high power 8Ω load at 2.5A. Case temperature was monitored and the test shut down when the top front case temperature reached 70° and was still increasing. 70° is sufficient to burn skin, and is an unsafe operating temperature.
A later manufacture EP-613 was put to the same test and it was still increasing in temperature at 70°. The vents in this supply allowed measurement of the heat sink temperature, and it was 110° which is very high.
Poor installation of the power transistors increases junction temperature.
θjc for a 2N3055 is 1.5°/W, and θch should be less than 0.5°/W, so at full load where each transistor dissipates 40W, the junction should be 80° higher than the heat sink temperature. The maximum safe junction temperature is 200°, so the heatsink must not be allowed to rise above 120°.
A poorly fitted transistor will increase θch and could contribute sufficient temperature rise to damage the transistors at lower power.
The power supply depends on natural convection cooling, and in the case of the MP-3082, the case vents are covered with a very fine mesh which further restricts air flow. The mesh will be removed from the top surface of the cover to assist air flow.
Inadequate air flow has contributed to excessive heat sink temperature and transistor damage.
A small fan will be fitted to assist air flow.
Considering the case of the power supply loaded by a short circuit and current limit set to the rated 2.5A, the total input power to the power supply with a 0Ω load at 2.5A is 78W. Since there is not output power under those conditions, the power supply must dissipate all 78W. (Note that current drawn from the 250V AC supply was 0.37A, so apparent power is 92.5VA, but power factor is 0.84 and real power is 78W.)
If we assume that the air passing through the fan must remove all of the all of the heat generated within the power supply, we have a conservative design guide, and if we assume that the metal case will be no hotter than the exhaust air and no higher than 60°, we can estimate the volume of air required for an air temperature rise of no more than 60-20=40°.
The heat capacity of dry air at room temperature is approximately 1.01J/g/K. So, in 1s, 78W delivers 78J of energy, and the mass of air that 78J that will raise 40° is 78*1.01/40=1.97g. The density of air at room temperature and pressure is about 1.2g/l, so the volume of air would be 1.97/1.2=1.64l. This is a flow of 1.64l/s.
In this application the exhaust vents are generous in comparison to the fan aperture, so let us assume that pressure will be very low, and that the fans rated maximum flow needs to be only 25% higher, or 2l/s (0.12m^3/min, 4.3CFM).
A candidate fan is the Ceramica 4010C12M-II ceramic spindle (for long life and low noise) low power 40mm muffin fans rated at 0.17m^3/min (2.8l/s). The fan requires 90mA at 12V, and will be powered from the unregulated supply for the 5V auxiliary regulator, which measures 13V with no load.
Fig 1 shows the characteristics of the proposed fan.
Considering still the case of the power supply loaded by a short circuit and current limit set to the rated 2.5A, the total input power to the power supply with a 0Ω load at 2.5A is 78W. Since there is not output power under those conditions, the power supply must dissipate all 78W. (Note that current drawn from the 250V AC supply was 0.37A, so apparent power is 92.5VA, but power factor is 0.84 and real power is 78W.)
If we assume that the air passing through the fan must remove all of the all of the heat generated within the power supply, we can estimate the temperature rise of the exhaust air.
The fan is rated for 2.8l/s (0.17m^3/min) at zero static pressure. In this application the exhaust vents are generous in comparison to the fan aperture, so let us assume that pressure will be low and 80% of that air flow will be achieved in this application, 2.24l/s.
The density of air at room temperature is about 1.2g/l, so the air flow represents about 2.7g/s.
The heat capacity of dry air at room temperature is approximately 1.01J/g/K. So 78W (78J/s) will raise 2.7g/s of air by 78/2.7/1.01=29K, so intake air at 20°C will be raised to 49°C and the exterior metal should not reach this temperature. In practice, some radiation loss may result in exterior metal reaching a slightly higher temperature, but it is unlikely to reach near 60°C (the maximum that is safe for body contact).
The candidate fans (Ceramica 4010C12M-II) have a three pin polarised female header connector fitted, so the matching male connector was used to provide an attachment point to the panel PCB. Only two pins are required for this application, the + and - 12V wires, so the unused pin was removed from the male connector.
Fig 2 shows the male connector attached to the rectifier for the auxiliary 5V supply. The front panel needs to be detached to get access for attachment of the connector.
Fig 3 shows the fan connection with the front panel reinstalled.
Fig 4 shows the case drilled for a handle and the fan. The holes were deburred and some black satin paint applied with a swap to the edges of the holes to prevent rust. Hint: drill the small holes before using a holesaw for the large hole. This case is the newer case with the unobstructed vent holes.
The mesh covering the vents on the top face of the MP-3082 was removed by cutting the spot welds. Done carefully, this will not damage the exterior paint, though again, a touch up of damaged paint on the interior will help to prevent rust.
Fig 5 shows a finished modified power supply. A chromed finger grill provides some protection against touching the fan.
The modified MP-3082 on test with a high power 8Ω load at 2.5A after 30min has stabilised, temperatures were:
Cost of parts for the modification total A$8.
A further important test is whether the power supply can withstand a short circuit on the load terminals with the current limit set to 2.5A (the rated current).
The modified MP-3082 on test with a 0Ω load at 2.5A after 30min has stabilised, temperatures were:
These measurements reconcile with the heat design set out earlier.
The subject MP-3802 failed on a load within specifications due to:
The power supply can be hardened to better handle its rated output by:
© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.